Image dataset processing method, apparatus, device, and storage medium

Abstract

Embodiments of the present disclosure provide an image dataset processing method, device and equipment and a storage medium. The method comprises: obtaining an initial autonomous driving image dataset; inputting the initial image dataset into a first autonomous driving detection model and a second autonomous driving detection model respectively, and obtaining a first image detection result and a second image detection result respectively; comparing the first image detection result and the second image detection result to obtain a comparison result; determining a comparison type according to the comparison result; determining evaluation scores of a plurality of initial autonomous driving images according to the comparison type; selecting a target autonomous driving image dataset from the initial autonomous driving image dataset according to the evaluation scores of the plurality of initial autonomous driving images, and adjusting the labels in the target autonomous driving image dataset. Embodiments of the present disclosure can improve the accuracy of image dataset processing.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of image processing, and particularly relate to an image dataset processing method and device, equipment and a storage medium. BACKGROUND

[0002] It is known that the development of automatic driving is closely related to the progress of artificial intelligence technology. Data, computing power and algorithms, as the three elements of artificial intelligence, significantly affect the maturity of automatic driving technology, and among the three elements, data plays a crucial role. Artificial intelligence can only summarize rules after a large amount of data training. In practical applications, if a scenario that has never occurred in the training set appears, the artificial intelligence is basically in a blind guessing state, resulting in prediction errors. Therefore, reliable and high-quality data is particularly important in the development process of automatic driving.

[0003] Although the data collected by a high-level automatic driving test vehicle every day is in terabytes (TB) level capacity, and needs petabytes (PB) level storage space, among these data, the valuable data that can be used for training automatic driving is less than or equal to about 5%. How to find the most valuable difficult samples in the massive data and prioritize the annotation of these difficult samples to accelerate the iteration of the automatic driving model and the mass production of algorithms is a problem that needs to be solved. SUMMARY

[0004] Embodiments of the present disclosure provide an image dataset processing method, device, equipment and storage medium, which can improve the accuracy of image dataset processing.

[0005] In a first aspect, embodiments of the present disclosure provide an image dataset processing method, comprising: obtaining an initial automatic driving image dataset; wherein the initial automatic driving image dataset comprises a plurality of initial automatic driving images and corresponding labels; inputting the initial image dataset into a first automatic driving detection model and a second automatic driving detection model respectively, to obtain a first image detection result and a second image detection result respectively; comparing the first image detection result and the second image detection result to obtain a comparison result; determining a comparison type according to the comparison result; wherein the first image detection result is taken as a reference detection result; the comparison type comprises true positive, false positive and false negative; determining an evaluation score of the plurality of initial automatic driving images according to the comparison type; and screening a target automatic driving image dataset from the initial automatic driving image dataset according to the evaluation score of the plurality of initial automatic driving images, and adjusting the labels in the target automatic driving image dataset.

[0006] In a second aspect, the embodiments of the present disclosure further provide an image dataset processing apparatus, including: an initial autonomous driving image dataset obtaining module, configured to obtain an initial autonomous driving image dataset; wherein the initial autonomous driving image dataset includes a plurality of initial autonomous driving images and corresponding labels; an image detection result obtaining module, configured to input the initial image dataset into a first autonomous driving detection model and a second autonomous driving detection model respectively, and obtain a first image detection result and a second image detection result respectively; an image detection result comparison module, configured to compare the first image detection result and the second image detection result, and obtain a comparison result; a comparison type determining module, configured to determine a comparison type according to the comparison result; wherein the first image detection result is taken as a reference detection result; the comparison type includes true positive, false positive and false negative; an evaluation score determining module, configured to determine evaluation scores of the plurality of initial autonomous driving images according to the comparison type; and a target autonomous driving image dataset screening module, configured to screen a target autonomous driving image dataset from the initial autonomous driving image dataset according to the evaluation scores of the plurality of initial autonomous driving images, and adjust labels in the target autonomous driving image dataset.

[0007] In a third aspect, the embodiments of the present disclosure further provide an electronic device, including:

[0008] one or more processors;

[0009] a memory device configured to store one or more programs,

[0010] when the one or more programs are executed by the one or more processors, the one or more processors implement the image dataset processing method according to the embodiments of the present disclosure.

[0011] In a fourth aspect, the embodiments of the present disclosure further provide a storage medium containing computer executable instructions, when the computer executable instructions are executed by a computer processor, the computer processor is used to execute the image dataset processing method according to the embodiments of the present disclosure.

[0012] The technical scheme disclosed by the embodiment comprises: obtaining an initial automatic driving image data set; wherein the initial automatic driving image data set comprises a plurality of initial automatic driving images and corresponding labels; inputting the initial image data set into a first automatic driving detection model and a second automatic driving detection model respectively to obtain a first image detection result and a second image detection result respectively; comparing the first image detection result and the second image detection result to obtain a comparison result; determining a comparison type according to the comparison result; wherein the first image detection result is taken as a reference detection result; the comparison type comprises a true positive example, a false positive example and a false negative example; determining evaluation scores of the plurality of initial automatic driving images according to the comparison type; and screening a target automatic driving image data set from the initial automatic driving image data set according to the evaluation scores of the plurality of initial automatic driving images, and adjusting the labels in the target automatic driving image data set. According to the embodiment, the comparison type is determined according to the comparison result between the first image detection result and the second image detection result, the evaluation scores of the initial automatic driving images are determined according to the comparison type, and the target automatic driving image data set is screened according to the evaluation scores of the initial automatic driving images, so that the accuracy of image data set processing can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0013] The above-described and other features and advantages of the various embodiments of the present disclosure will be more apparent from the following detailed description taken in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals are used to represent the same or similar elements. It should be understood that the drawings are schematic and elements and features are not necessarily to scale.

[0014] Figure 1 A flowchart of an image data set processing method provided by an embodiment of the present disclosure is shown in FIG. 1.

[0015] Figure 2 A flowchart of another image data set processing method provided by an embodiment of the present disclosure is shown in FIG. 2.

[0016] Figure 3 An effect diagram of an intersection over union and an intersection over union evaluation score provided by an embodiment of the present disclosure is shown in FIG. 3.

[0017] Figure 4 An effect diagram of a confidence and a second evaluation score provided by an embodiment of the present disclosure is shown in FIG. 4.

[0018] Figure 5 An effect diagram of a confidence and a first evaluation score provided by an embodiment of the present disclosure is shown in FIG. 5.

[0019] Figure 6 An effect diagram of an area and a second evaluation score provided by an embodiment of the present disclosure is shown in FIG. 6.

[0020] Figure 7An effect diagram between the area and the first evaluation score provided for the embodiment of the present application;

[0021] Figure 8 An effect diagram of the first image detection result provided for the embodiment of the present application;

[0022] Figure 9 An effect diagram of the second image detection result provided for the embodiment of the present application;

[0023] Figure 10 A structure diagram of an image data set processing device provided by the embodiment of the present disclosure;

[0024] Figure 11 A structure diagram of an electronic device provided by the embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] Embodiments of the present disclosure will be described in more detail by referring to the drawings. Although certain embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms, and should not be interpreted as being limited to the embodiments set forth herein, but rather, these embodiments are provided to more thoroughly and completely understand the present disclosure. It is understood that the drawings and embodiments of the present disclosure are for exemplary purposes only, and are not intended to limit the scope of protection of the present disclosure.

[0026] It is understood that each step recited in the method embodiment of the present disclosure can be executed in different order, and / or in parallel. In addition, the method embodiment can include additional steps and / or omit the execution of the steps shown. The scope of the present disclosure is not limited in this respect.

[0027] The term "comprising" and variations thereof as used herein are open-ended, that is "including, but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Related definitions are given below in the description of the various terms.

[0028] It should be noted that the concepts of "first", "second", etc. mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0029] It should be noted that the modification of "one", "multiple" mentioned in the present disclosure is illustrative and not limiting, and those skilled in the art should understand that, unless otherwise explicitly indicated in the context, it should be understood as "one or more".

[0030] It can be understood that the data involved in the technical solution (including but not limited to the data itself, the acquisition or use of the data) should comply with the requirements of the corresponding laws, regulations and relevant provisions.

[0031] Figure 1 A flowchart of an image data set processing method is provided for an embodiment of the present application. The present embodiment can be applicable to the case of screening a target autonomous driving image data set from an initial autonomous driving image data set. It should be noted that the present embodiment is not limited to the task of scene target detection of the front-view camera of autonomous driving, but can also be other scene tasks. When switching scene tasks, the relevant configuration information is modified according to the model capability and project requirements. In the present embodiment, the case of screening a target autonomous driving image data set from an initial autonomous driving image data set can also be understood as a data mining case, so as to find the most valuable difficult samples in a large amount of data and prioritize annotation to accelerate model iteration. The method can be executed by an image data set processing device, and specifically includes the following steps:

[0032] S110, acquiring an initial autonomous driving image data set.

[0033] The initial autonomous driving image data set includes a plurality of initial autonomous driving images and corresponding annotations. The annotations include detection box information, class, confidence, area, etc. of the initial autonomous driving images. When the present embodiment is used for a 3-dimensional scene task, the area can be a 3-dimensional target volume or a projected area.

[0034] The initial autonomous driving images can be collected by a high-level autonomous driving test vehicle. The autonomous driving image data set can or can not include annotations, and the present embodiment does not limit this.

[0035] S120, inputting the initial image data set into a first autonomous driving detection model and a second autonomous driving detection model respectively to obtain first image detection results and second image detection results respectively.

[0036] The first autonomous driving detection model can be a server model, which has sufficient computing power and can be processed offline, and can be referred to as a large-scale model. The second autonomous driving detection model can be an edge model, which has limited computing power and high real-time requirements, and can be referred to as a small-scale model.

[0037] Optionally, each of the initial automatic driving images comprises at least one target object; the first image detection result comprises first bounding box information of the at least one target object, first categories corresponding to the at least one target object respectively, first confidence degrees of the first categories, and first areas corresponding to the first bounding box information; and the second image detection result comprises second bounding box information of the at least one target object, second categories corresponding to the at least one target object respectively, second confidence degrees corresponding to the second categories, and second areas corresponding to the second bounding box information.

[0038] The target object can be a vehicle, a railing, a pedestrian, a house, etc., and the embodiment is not limited in this regard. The bounding box information can comprise a length, a width, and a height of the bounding box.

[0039] S130, comparing the first image detection result and the second image detection result to obtain a comparison result.

[0040] In the embodiment, the first image detection result and the second image detection result can be compared to obtain a corresponding comparison result. The comparison result comprises a consistent comparison and an inconsistent comparison.

[0041] Optionally, the comparison result comprises a category comparison result. The comparison of the first image detection result and the second image detection result to obtain the comparison result comprises: comparing the first categories and the second categories to obtain the category comparison result; and the category comparison result comprises a category consistent comparison and a category inconsistent comparison.

[0042] In the embodiment, the category comparison result can be obtained by comparing the first categories and the second categories, and the comparison type can be determined according to the category comparison result.

[0043] S140, determining a comparison type according to the comparison result.

[0044] The first image detection result is taken as a reference detection result; and the comparison type comprises a true positive (TP), a false positive (FP), and a false negative (FN). For example, if the category comparison is consistent, the comparison type is a true positive; if the category comparison is inconsistent, the comparison type is a false positive or a false negative.

[0045] Optionally, the comparison type is determined according to the comparison result, including: determining an intersection-over-union according to the first bounding box information and the second bounding box information; determining a first intersection-over-union set value according to the first area and / or the second area; if the intersection-over-union is greater than or equal to the first intersection-over-union set value, and the category comparison result is category comparison consistent, then the comparison type is determined as a true example; if the intersection-over-union is less than the first intersection-over-union set value and / or the category comparison result is category comparison inconsistent, then the comparison type is determined as a false positive and a false negative; or; the first image detection result includes the first bounding box information of the target object, and the second image detection result does not include the second bounding box information of the target object, then the comparison type is determined as a false negative; the first image detection result does not include the first bounding box information of the target object, and the second image detection result includes the second bounding box information of the target object, then the comparison type is determined as a false positive.

[0046] In this embodiment, the intersection-over-union can be determined according to the first bounding box information and the second bounding box information; the size of the target object can be determined according to the first area and / or the second area, if the first area and / or the second area is less than 32*32, then the target object is a small target, and the corresponding first intersection-over-union set value can be 0.3, if the first area and / or the second area is greater than or equal to 32*32, then the target object is a normal target, and the corresponding first intersection-over-union set value can be 0.5.

[0047] If the intersection-over-union is greater than or equal to the first intersection-over-union set value, and the category comparison result is category comparison consistent, then the comparison type is determined as a true example. If the intersection-over-union is less than the first intersection-over-union set value and / or the category comparison result is category comparison inconsistent, then the comparison type is determined as a false positive and a false negative. Or; the first image detection result includes the first bounding box information of the target object, and the second image detection result does not include the second bounding box information of the target object, then the comparison type is determined as a false negative, that is, the detection box in the first image detection result relative to the second image detection result is a false negative. The first image detection result does not include the first bounding box information of the target object, and the second image detection result includes the second bounding box information of the target object, then the comparison type is determined as a false positive, that is, the detection box in the second image detection result relative to the first image detection result is a false positive.

[0048] In this embodiment, the comparison type is determined according to the comparison result and the intersection-over-union, or the comparison type is determined according to whether the first image detection result and the second image detection result include the same detection box information of the target object, which can accurately determine the comparison type.

[0049] S150, determining the evaluation score of the plurality of initial automatic driving images according to the comparison type.

[0050] In this embodiment, the evaluation score of each initial automatic driving image is determined as follows: the evaluation scores of all target objects in the initial automatic driving image are determined according to the comparison type, the evaluation scores of the target objects include the intersection-over-union score, the confidence score, the area score and the category score, and the evaluation score of the initial automatic driving image is obtained according to the evaluation scores of all target objects and the scene score of the initial automatic driving image, so that the evaluation scores of all initial automatic driving images can be obtained.

[0051] In S160, the target automatic driving image dataset is screened from the initial automatic driving image dataset according to the evaluation scores of the plurality of initial automatic driving images, and the labeling in the target automatic driving image dataset is adjusted.

[0052] In this embodiment, the target automatic driving image dataset can be screened from the initial automatic driving image dataset according to the evaluation scores of all initial automatic driving images. The target automatic driving image dataset can be understood as a dataset that has a greater impact on the detection accuracy of the second automatic driving detection model (or other automatic driving detection models), and can also be understood as a dataset with a greater target value. By adjusting the labeling in the target automatic driving image dataset, the detection accuracy of the automatic driving detection model can be improved.

[0053] In this embodiment, the comparison is not limited to between two models, but can also be between two models among a plurality of models. The evaluation score of one initial automatic driving image can obtain a plurality of value scores, and the maximum score or the final value score is obtained by other fusion score algorithms.

[0054] In this embodiment, only one model can be used, but two different modes of forward calculation need to be performed, such as TTA (Test Time Augmentation) mode and normal mode, to obtain two different detection results.

[0055] The technical scheme disclosed by the embodiment comprises the following steps: obtaining an initial automatic driving image dataset; wherein the initial automatic driving image dataset comprises a plurality of initial automatic driving images and corresponding labels; inputting the initial image dataset into a first automatic driving detection model and a second automatic driving detection model respectively to obtain a first image detection result and a second image detection result respectively; comparing the first image detection result and the second image detection result to obtain a comparison result; determining a comparison type according to the comparison result; wherein the first image detection result serves as a reference detection result; the comparison type comprises a true positive example, a false positive example and a false negative example; determining an evaluation score of the plurality of initial automatic driving images according to the comparison type; and screening a target automatic driving image dataset from the initial automatic driving image dataset according to the evaluation score of the plurality of initial automatic driving images and adjusting the labels in the target automatic driving image dataset. According to the method, the accuracy of image dataset processing can be improved.

[0056] Figure 2 The flowchart of another image dataset processing method provided by the embodiment is shown in FIG. 2. The embodiment is a specific implementation based on the above-mentioned embodiment. The method provided by the embodiment comprises the following steps: Figure 2

[0057] S201, obtaining an initial automatic driving image dataset.

[0058] S202, inputting the initial image dataset into a first automatic driving detection model and a second automatic driving detection model respectively to obtain a first image detection result and a second image detection result respectively.

[0059] S203, comparing the first image detection result and the second image detection result to obtain a comparison result.

[0060] S204, determining a comparison type according to the comparison result.

[0061] S205, determining an evaluation score of a target object in the initial automatic driving image according to the comparison type for the evaluation score of any one initial automatic driving image.

[0062] The evaluation score of the target object comprises an intersection over union score, a confidence score, an area score and a category score of the target object. In the embodiment, the evaluation score of the target object is different for different comparison types. According to the comparison type, the evaluation score of all target objects or the evaluation score of part of the target objects in the initial automatic driving image can be determined. ​

[0063] Optionally, the evaluation score of the target object in the initial automatic driving image is determined according to the comparison type, including: determining at least one of the intersection-over-union score, the confidence score and the area score of the target object according to the comparison type; obtaining a pre-set category score corresponding to the target object; and determining the evaluation score of the target object according to at least one of the intersection-over-union score, the confidence score, the area score and the category score.

[0064] In the embodiment, the evaluation score of each target object can be obtained by: obtaining the intersection-over-union score, the confidence score and the area score of the target object according to the comparison type respectively, and obtaining a pre-set category score corresponding to the target object. The intersection-over-union score, the confidence score, the area score and the category score are multiplied or weightedly averaged, and the result of multiplication or the result of weighted average is taken as the evaluation score of the target object. In the embodiment, the elements for determining the evaluation score of the target object are not limited to the intersection-over-union score, the confidence score, the area score and the category score, and the corresponding elements can be increased or reduced.

[0065] It should be noted that the category weight corresponding to each target object is pre-set, that is, the category score, and the category score ranges from 0 to 1. After predicting the category to which the target object belongs, the category score corresponding to the predicted category can be obtained. For the category weight, the initial automatic driving image data set distribution information can be fused, and higher category weight is set for the long tail part. The evaluation result of the model on the test set can also be fused, and the weight is increased for the category with poor accuracy.

[0066] In the embodiment, the evaluation score of the target object is determined according to at least one of the intersection-over-union score, the confidence score, the area score and the category score, so that the evaluation score of the target object can be accurately determined.

[0067] Optionally, the intersection-over-union score of the target object is determined according to the comparison type, including: determining a second intersection-over-union setting value according to the first area and / or the second area; and determining the intersection-over-union evaluation score according to the comparison type, the first intersection-over-union setting value, the second intersection-over-union setting value and the intersection-over-union.

[0068] The second intersection-over-union setting value is greater than the first intersection-over-union setting value; the second intersection-over-union setting value can be understood as a maximum intersection-over-union threshold, and the first intersection-over-union setting value can be understood as a minimum intersection-over-union threshold. In the embodiment, the size of the target object is determined according to the first area and / or the second area. If the first area and / or the second area is less than 32*32, the target object is a small target, and the corresponding second intersection-over-union setting value can be 0.7. If the first area and / or the second area is greater than or equal to 32*32, the target object is a normal target, and the corresponding first intersection-over-union setting value can be 0.9.

[0069] An exemplary intersection over union threshold formula is as follows:

[0070]

[0071] where iou_thresh represents the intersection over union threshold, the first intersection over union setting value can be 0.3 and the second intersection over union setting value can be 0.7 when the target object is a small target. The first intersection over union setting value can be 0.5 and the second intersection over union setting value can be 0.9 when the target object is a normal target.

[0072] Specifically, if the comparison type is false positive or false negative and the intersection over union is less than the first intersection over union setting value, the intersection over union evaluation score can be directly obtained. If the comparison type is true positive and the intersection over union is greater than or equal to the second intersection over union setting value, the intersection over union evaluation score can be directly obtained. If the comparison type is true positive and the intersection over union falls within the interval formed by the first intersection over union setting value and the second intersection over union setting value, the intersection over union evaluation score can be determined according to the first intersection over union setting value, the second intersection over union setting value, and the intersection over union.

[0073] In this embodiment, the intersection over union evaluation score is determined by the comparison type, the first intersection over union setting value, the second intersection over union setting value, and the intersection over union, which can accurately determine the intersection over union evaluation score.

[0074] Optionally, determining the intersection over union evaluation score according to the comparison type, the first intersection over union setting value, the second intersection over union setting value, and the intersection over union includes: if the comparison type is false positive or false negative and the intersection over union is less than the first intersection over union setting value, the intersection over union evaluation score is the first set intersection over union evaluation score; if the comparison type is true positive and the intersection over union is greater than or equal to the second intersection over union setting value, the intersection over union evaluation score is the second set intersection over union evaluation score; if the comparison type is true positive and the intersection over union falls within the interval formed by the first intersection over union setting value and the second intersection over union setting value, the first intersection over union evaluation score is determined according to the second intersection over union setting value and the intersection over union; the second intersection over union evaluation score is determined according to the second intersection over union setting value and the first intersection over union setting value; the intersection over union evaluation score is determined according to the first intersection over union evaluation score and the second intersection over union evaluation score.

[0075] An exemplary intersection over union evaluation score formula is as follows:

[0076]

[0077] where iou_score is the intersection over union evaluation score, min_iou is the first intersection over union setting value, max_iou is the second intersection over union setting value, and iou is the intersection over union. The first set intersection over union evaluation score is 1 and the second set intersection over union evaluation score is 0.

[0078] wherein, in the case of the comparison type being false positive or false negative and the intersection over union being less than the first intersection over union setting value, the value is the highest, the intersection over union evaluation score is 1; in the case of the comparison type being true positive and the intersection over union being greater than or equal to the second intersection over union setting value, it can be considered that the model does not need to pay attention to the small edge difference, and there is no value, the intersection over union evaluation score is 0; in the case of the comparison type being true positive and the intersection over union falling within the interval formed by the first intersection over union setting value and the second intersection over union setting value, the lower the intersection over union, the higher the value, the first intersection over union evaluation score is max_iou-iou, the second intersection over union evaluation score is max_iou-min_iou, and the intersection over union evaluation score is

[0079] Exemplarily, Figure 3 An effect diagram of the intersection over union and the intersection over union evaluation score provided by the embodiment of the present application is shown in FIG. 1. As shown in the figure, the horizontal coordinate is the intersection over union iou, and the vertical coordinate is the intersection over union evaluation score iou_score. It can be seen that the target object is a small target or a normal target, and the intersection over union and the intersection over union evaluation score are inversely proportional. Figure 3

[0080] It should be noted that the target object is treated differently according to the target size because the absolute coordinate deviation (measured in pixels) of the same target position has a greater impact on the iou value of a small target during registration, so the iou value of a general small target is not very high, which will make the evaluation score of more small targets higher, which is not consistent with the concept that “large targets have greater value”.

[0081] In the embodiment, in the case of the comparison type being false positive or false negative and the intersection over union being less than the first intersection over union setting value, the first setting intersection over union evaluation score is taken as the intersection over union evaluation score; in the case of the comparison type being true positive and the intersection over union being greater than or equal to the second intersection over union setting value, the second setting intersection over union evaluation score is taken as the intersection over union evaluation score; and in the case of the comparison type being true positive and the intersection over union falling within the interval formed by the first intersection over union setting value and the second intersection over union setting value, the intersection over union evaluation score is determined according to the first intersection over union evaluation score and the second intersection over union evaluation score, so that the intersection over union evaluation score can be accurately determined.

[0082] ​Optionally, the confidence score of the target object is determined according to the comparison type, including: if the comparison type is false positive, a first evaluation score of the second confidence is determined, and the first evaluation score of the second confidence is taken as the confidence score of the target object; if the comparison type is false negative, a first evaluation score of the first confidence is determined, and the first evaluation score of the first confidence is taken as the confidence score of the target object; and if the comparison type is true positive, a second evaluation score of the first confidence and a second evaluation score of the second confidence are respectively determined, and the confidence score of the target object is determined according to the second evaluation score of the first confidence and the second evaluation score of the second confidence.

[0083] For example, the confidence score of the target object is determined according to the following formula:

[0084]

[0085] wherein, matched_conf_score is the confidence score of the target object, conf_score small1 is the first evaluation score of the second confidence when the comparison type is false positive, that is, the confidence score of the target object when the comparison type is false positive. conf_score big1 is the first evaluation score of the first confidence when the comparison type is false negative, that is, the confidence score of the target object when the comparison type is false negative. is the confidence score of the target object when the comparison type is true positive. conf_score big2 is the second evaluation score of the first confidence, conf_score small2 is the second evaluation score of the second confidence.

[0086] In this embodiment, if the comparison type is false positive, the first evaluation score of the second confidence output by the second automatic driving detection model is determined, and the first evaluation score of the second confidence is taken as the confidence score of the target object when the comparison type is false positive; if the comparison type is false negative, the first evaluation score of the first confidence output by the first automatic driving detection model is determined, and the first evaluation score of the first confidence is taken as the confidence score of the target object when the comparison type is false negative; and if the comparison type is true positive, the second evaluation score of the first confidence and the second evaluation score of the second confidence are respectively determined, and the confidence score of the target object when the comparison type is true positive is determined according to the second evaluation score of the first confidence and the second evaluation score of the second confidence, so that the confidence score of the target object can be accurately determined.

[0087] Optionally, the first evaluation score of the first confidence level or the first evaluation score of the second confidence level is determined by: obtaining a first confidence level threshold and a second confidence level threshold; wherein the second confidence level threshold is greater than the first confidence level threshold; if the first confidence level or the second confidence level is less than the first confidence level threshold, a first set confidence level score is taken as the first evaluation score of the first confidence level or the first evaluation score of the second confidence level; if the first confidence level is greater than or equal to the second confidence level threshold or the second confidence level is greater than or equal to the second confidence level threshold, a second set confidence level score is taken as the first evaluation score of the first confidence level or the first evaluation score of the second confidence level; if the first confidence level falls within an interval formed by the first confidence level threshold and the second confidence level threshold, a first confidence level score is determined according to the first confidence level, the second confidence level threshold and the first confidence level threshold; a second confidence level score is determined according to the second confidence level threshold and the first confidence level threshold; the first evaluation score of the first confidence level is determined according to the first confidence level score and the second confidence level score; or if the second confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, a third confidence level score is determined according to the second confidence level, the second confidence level threshold and the first confidence level threshold; a fourth confidence level score is determined according to the second confidence level threshold and the first confidence level threshold; the first evaluation score of the second confidence level is determined according to the third confidence level score and the fourth confidence level score.

[0088] For example, in the case of the comparison type being FP, the first evaluation score of the second confidence level is determined according to the following formula:

[0089]

[0090] The second confidence level threshold can be 0.5 and the first confidence level threshold can be 0.3. The first set confidence level score can be 0 and the second set confidence level score can be 1.

[0091] In the case of the comparison type being false positive, the conf is the second confidence level; if the second confidence level is less than the first confidence level threshold, the first evaluation score of the second confidence level is 0, which is of no value; if the second confidence level is greater than or equal to the second confidence level threshold, the first evaluation score of the second confidence level is 1, which is of the greatest value; if the second confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, is the first evaluation score of the second confidence level.

[0092] In the embodiment, if the second confidence is less than the first confidence threshold, the first set confidence score is taken as the first evaluation score of the second confidence; if the second confidence is greater than or equal to the second confidence threshold, the second set confidence score is taken as the first evaluation score of the second confidence; and if the second confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, the first evaluation score of the second confidence is determined according to the third confidence score and the fourth confidence score, so that the first evaluation score of the second confidence can be accurately determined.

[0093] For example, in the case of the comparison type FN, the formula for determining the first evaluation score of the first confidence is as follows:

[0094]

[0095] When the comparison type is false negative, conf is the first confidence; if the first confidence is less than the first confidence threshold, the first evaluation score of the first confidence is 0, which is of no value; if the first confidence is greater than or equal to the second confidence threshold, the first evaluation score of the first confidence is 1, which is of the greatest value; and if the first confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, the first evaluation score of the first confidence is determined according to the third confidence score and the fourth confidence score.

[0096] In the embodiment, if the first confidence is less than the first confidence threshold, the first set confidence score is taken as the first evaluation score of the first confidence; if the first confidence is greater than or equal to the second confidence threshold, the second set confidence score is taken as the first evaluation score of the first confidence; and if the first confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, the first evaluation score of the first confidence is determined according to the first confidence score and the second confidence score, so that the first evaluation score of the first confidence can be accurately determined.

[0097] ​Optionally, the second evaluation score of the first confidence level is determined, or the second evaluation score of the second confidence level is determined, including: if the first confidence level or the second confidence level is less than the first confidence level threshold, the first set confidence level score is taken as the second evaluation score of the first confidence level or the second evaluation score of the second confidence level; if the first confidence level is greater than or equal to the first confidence level threshold, or the second confidence level is greater than or equal to the second confidence level threshold, the first set confidence level score is taken as the second evaluation score of the first confidence level or the second evaluation score of the second confidence level; if the first confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, a fifth confidence level score is determined according to the first confidence level and the second confidence level threshold; a sixth confidence level score is determined according to the second confidence level threshold and the first confidence level threshold; the second evaluation score of the first confidence level is determined according to the fifth confidence level score and the sixth confidence level score; or if the second confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, a seventh confidence level score is determined according to the second confidence level and the second confidence level threshold; an eighth confidence level score is determined according to the second confidence level threshold and the first confidence level threshold; the second evaluation score of the second confidence level is determined according to the seventh confidence level score and the eighth confidence level score.

[0098] Exemplarily, in the case of the comparison type being a true example, the formula for determining the second evaluation score of the second confidence level is as follows:

[0099]

[0100] In the case of the comparison type being a true example, conf1 is the second confidence level; if the second confidence level is less than the first confidence level threshold, the second evaluation score of the second confidence level is 0, which is worthless; if the second confidence level is greater than or equal to the second confidence level threshold, the second evaluation score of the second confidence level is 0, which is worthless; if the second confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, is the second evaluation score of the second confidence level.

[0101] In this embodiment, the second evaluation score of the second confidence level is accurately determined by the way that if the second confidence level is less than the first confidence level threshold, the first set confidence level score is taken as the second evaluation score of the second confidence level; if the second confidence level is greater than or equal to the second confidence level threshold, the first set confidence level score is taken as the second evaluation score of the second confidence level; if the second confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, the second evaluation score of the second confidence level is determined according to the seventh confidence level score and the eighth confidence level score.

[0102] Exemplarily, in the case of the comparison type being a true example, the formula for determining the second evaluation score of the first confidence level is as follows:

[0103]

[0104] In the case of a true instance comparison, conf2 represents the first confidence level. If the first confidence level is less than the first confidence threshold, the second evaluation score of the first confidence level is 0. If the first confidence level is greater than or equal to the first confidence threshold, the second evaluation score of the first confidence level is 0. If the first confidence level falls within the interval formed by the first and second confidence thresholds, then... The second assessment score is the score for the first confidence level.

[0105] In this embodiment, if the first confidence level is less than the first confidence level threshold, then the first set confidence level score is used as the second evaluation score of the first confidence level; if the first confidence level is greater than or equal to the first confidence level threshold, then the first set confidence level score is used as the second evaluation score of the first confidence level; if the first confidence level falls within the interval formed by the first confidence level threshold and the second confidence level threshold, then the second evaluation score of the first confidence level is determined based on the fifth confidence level score and the sixth confidence level score. This method can accurately determine the second evaluation score of the first confidence level.

[0106] For example, Figure 4 This is a schematic diagram illustrating the relationship between confidence level and second evaluation score, provided as an embodiment of the present invention. Figure 4 As shown, when the comparison type is a true instance, the horizontal axis represents `conf`, where `conf` is the first or second confidence level, `min_conf` is the first confidence threshold, and `max_conf` is the second confidence threshold. The vertical axis represents `conf_score`, where `conf_score` is the second evaluation score for the first or second confidence level. When the first or second confidence level is less than the first confidence threshold, or greater than or equal to the second confidence threshold, the second evaluation score is 0, indicating no value. When `conf` is between the first and second confidence thresholds, the second evaluation score (which can also be understood as value) is inversely proportional to the confidence level.

[0107] For example, Figure 5 This is a schematic diagram illustrating the relationship between confidence level and first evaluation score, provided as an embodiment of the present invention. Figure 5As shown, in case of false negative or false positive, the horizontal axis is conf, the first confidence or the second confidence, min_conf is the first confidence threshold, max_conf is the second confidence threshold, and the vertical axis is conf_score, the first evaluation score of the first confidence or the second confidence. When the first confidence or the second confidence is less than the first confidence threshold, the first evaluation score is all 0, which is worthless. When the first confidence or the second confidence is greater than or equal to the second confidence threshold, the value is the largest, and the first evaluation score is all 1. When the conf is between the first confidence threshold and the second confidence threshold, the first evaluation score (which can also be understood as the value) is proportional to the confidence.

[0108] Optionally, the area score of the target object is determined according to the comparison type, including: if the comparison type is false positive, the first evaluation score of the second area is determined, and the first evaluation score of the second area is taken as the area score of the target object; if the comparison type is false negative, the first evaluation score of the first area is determined, and the first evaluation score of the first area is taken as the area score of the target object; if the comparison type is true positive, the second evaluation score of the first area and the second evaluation score of the second area are determined respectively, and the area score of the target object is determined according to the second evaluation score of the first area and the second evaluation score of the second area.

[0109] For example, the area score of the target object is determined according to the following formula:

[0110]

[0111] Wherein, matched_area_score is the area score of the target object, area_score small1 is the first evaluation score of the second area in case of false positive, that is, the area score of the target object in case of false positive. area_score big1 is the first evaluation score of the first area in case of false negative, that is, the area score of the target object in case of false negative. is the area score of the target object in case of true positive. area_score big2 is the second evaluation score of the first area, area_score small2 is the second evaluation score of the second area.

[0112] In the embodiment, if the comparison type is false positive, the first evaluation score of the second area is determined as the area score of the target object in the case of false positive comparison type; if the comparison type is false negative, the first evaluation score of the first area is determined as the area score of the target object in the case of false negative comparison type; and if the comparison type is true positive, the area score of the target object in the case of true positive comparison type is determined according to the second evaluation score of the first area and the second evaluation score of the second area, so that the area score of the target object can be accurately determined.

[0113] Optionally, the first evaluation score of the first area or the first evaluation score of the second area is determined by: obtaining a first area threshold and a second area threshold; the second area threshold is greater than the first area threshold; if the first area or the second area is less than the first area threshold, a first set area score is taken as the first evaluation score of the first area or the first evaluation score of the second area; if the first area is greater than or equal to the second area threshold or the second area is greater than or equal to the second area threshold, a second set area score is taken as the first evaluation score of the first area or the first evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and the second area threshold, a first area score is determined according to the first area, the second area threshold and the first area threshold; a second area score is determined according to the second area threshold and the first area threshold; the first evaluation score of the first area is determined according to the first area score and the second area score; or if the second area falls within the interval formed by the first area threshold and the second area threshold, a third area score is determined according to the second area, the second area threshold and the first area threshold; a fourth area score is determined according to the second area threshold and the first area threshold; the first evaluation score of the second area is determined according to the third area score and the fourth area score.

[0114] The first area threshold can be 30*30 and the second area threshold can be 500*500; the first set area score can be 0; and the second set area score can be 1.

[0115] For example, in the case of FN comparison type, the formula for determining the first evaluation score of the first area is as follows:

[0116]

[0117] When the comparison type is false negative, the area is the first area, if the first area is less than the first area threshold, the first evaluation score of the first area is 0; if the first area is greater than or equal to the second area threshold, the first evaluation score of the first area is 1; if the first area falls within the interval formed by the first area threshold and the second area threshold, the first evaluation score of the first area is determined according to the first area, the second area threshold and the first area threshold. a first evaluation score of the first area is determined.

[0118] In the embodiment, if the first area is less than the first area threshold, the first set area score is taken as the first evaluation score of the first area; if the first area is greater than or equal to the second area threshold, the second set area score is taken as the first evaluation score of the first area; and if the first area falls within the interval formed by the first area threshold and the second area threshold, the first evaluation score of the first area is determined according to the first area score and the second area score.

[0119] For example, in the case of the comparison type being FP, the formula for determining the first evaluation score of the second area is as follows:

[0120]

[0121] wherein, when the comparison type is false positive, the area is the second area; if the second area is less than the first area threshold, the first evaluation score of the second area is 0; if the second area is greater than or equal to the second area threshold, the first evaluation score of the second area is 1; and if the second area falls within the interval formed by the first area threshold and the second area threshold, the first evaluation score of the second area is determined according to the third area score and the fourth area score. a first evaluation score of the second area is determined.

[0122] In the embodiment, if the second area is less than the first area threshold, the first set area score is taken as the first evaluation score of the second area; if the second area is greater than or equal to the second area threshold, the second set area score is taken as the first evaluation score of the second area; and if the second area falls within the interval formed by the first area threshold and the second area threshold, the first evaluation score of the second area is determined according to the third area score and the fourth area score.

[0123] Optionally, the second evaluation score of the first area is determined, or the second evaluation score of the second area is determined, including: if the first area or the second area is less than a first area threshold, taking the first set area score as the second evaluation score of the first area or the second evaluation score of the second area; if the first area is greater than or equal to the first area threshold, or if the second area is greater than or equal to the first area threshold, taking the second set area score as the second evaluation score of the first area or the second evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and a second area value, determining a fifth area score according to the first area and the first area threshold; determining a sixth area score according to the second area threshold and the first area threshold; determining the second evaluation score of the first area according to the fifth area score and the sixth area score; or if the second area falls within the interval formed by the first area threshold and the second area value, determining a seventh area score according to the second area and the first area threshold; determining an eighth area score according to the second area threshold and the first area threshold; determining the second evaluation score of the second area according to the seventh area score and the eighth area score.

[0124] Exemplarily, in the case of the comparison type being a true example, the formula for determining the second evaluation score of the first area is as follows:

[0125]

[0126] In the case of the comparison type being a true example, conf2 is the first area; if the first area is less than a first area threshold, the second evaluation score of the first area is 0; if the first area is greater than or equal to the first area threshold, the second evaluation score of the first area is 1; if the first area falls within an interval formed by the first area threshold and a second area value, the second evaluation score of the first area is determined according to the fifth area score and the sixth area score.

[0127] In this embodiment, by taking the first set area score as the second evaluation score of the first area if the first area is less than the first area threshold, taking the second set area score as the second evaluation score of the first area if the first area is greater than or equal to the first area threshold, and determining the second evaluation score of the first area according to the fifth area score and the sixth area score if the first area falls within the interval formed by the first area threshold and the second area value, the second evaluation score of the first area can be accurately determined.

[0128] Exemplarily, in the case of the comparison type being a true example, the formula for determining the second evaluation score of the second area is as follows:

[0129]

[0130] ​In the case of a true example, area2 is the second area; if the second area is less than the first area threshold, the second evaluation score of the second area is 0; if the second area is greater than or equal to the first area threshold, the second evaluation score of the second area is 1; if the second area falls within the interval formed by the first area threshold and the second area value, the second evaluation score of the second area is determined according to the seventh area score and the eighth area score. The second evaluation score of the second area is determined.

[0131] In the embodiment, if the second area is less than the first area threshold, the first set area score is taken as the second evaluation score of the second area; if the second area is greater than or equal to the first area threshold, the second set area score is taken as the second evaluation score of the second area; and if the second area falls within the interval formed by the first area threshold and the second area value, the second evaluation score of the second area is determined according to the seventh area score and the eighth area score, so that the second evaluation score of the second area can be accurately determined.

[0132] An exemplary effect diagram of the area and the second evaluation score is shown in FIG. 2. Figure 6 An exemplary effect diagram of the area and the first evaluation score is shown in FIG. 3. Figure 6 As shown in FIG. 2, in the case of a true example, the abscissa is area, the area is the first area or the second area, min_area is the first area threshold, and max_conf is the second area threshold. The ordinate is area_score, which is the second evaluation score. When the first area or the second area is less than the first area threshold, the second evaluation score is 0, which has no value. When the first area or the second area is greater than or equal to the second area threshold, the value is the largest, and the second evaluation score is 1. When the area is between the first area threshold and the second area threshold, the second evaluation score (which can also be understood as the value) is proportional to the area.

[0133] An exemplary effect diagram of the area and the first evaluation score is shown in FIG. 3. Figure 7 An exemplary effect diagram of the area and the first evaluation score is shown in FIG. 3. Figure 7 As shown in FIG. 3, in the case of a false negative example or a false positive example, the abscissa is area, the area is the first area or the second area, min_area is the first area threshold, max_area is the second area threshold, and the ordinate is area_score, which is the first evaluation score of the area. When the first area or the second area is less than the first area threshold, the first evaluation score is 0, which has no value. When the first area or the second area is greater than or equal to the second area threshold, the value is the largest, and the first evaluation score is 1. When the area is between the first area threshold and the second area threshold, the first evaluation score (which can also be understood as the value) is proportional to the area.

[0134] S206, determine a scene score of the initial autonomous driving image.

[0135] In this embodiment, the scene of the initial autonomous driving image is not limited, such as a rainy scene, a snowy scene, a traffic jam scene, an off-road scene, etc.

[0136] In this embodiment, the actual scene information can be obtained according to the initial autonomous driving image, the initial autonomous driving image is input into the autonomous driving scene model, the predicted scene information is output, and the scene score is obtained according to the similarity of the predicted scene information and the actual scene information.

[0137] Optionally, the scene score of the initial autonomous driving image is determined, comprising: inputting the initial autonomous driving image into the autonomous driving scene model, outputting the predicted scene information; obtaining the set scene information corresponding to the initial autonomous driving image; determining the scene similarity of the predicted scene information and the set scene information; determining the scene score of the initial autonomous driving image according to the scene similarity.

[0138] Wherein, the autonomous driving scene model can be a scene model based on any deep learning algorithm. The set scene information can be the real scene information corresponding to the initial autonomous driving image. In this embodiment, the initial autonomous driving image is input into the autonomous driving scene model, and the corresponding predicted scene information is output; the set scene information corresponding to the initial autonomous driving image is obtained, and the scene similarity between the predicted scene information and the set scene information is calculated based on any similarity algorithm; the scene score of the initial autonomous driving image is determined according to the scene similarity. Wherein, the range of scene similarity is [0, 1], that is, the range of scene score is [0, 1].

[0139] In this embodiment, the scene score of the initial autonomous driving image is determined by the scene similarity between the predicted scene information and the set scene information, which can accurately determine the scene score of the initial autonomous driving image.

[0140] S207, determine the evaluation score of the initial autonomous driving image according to the evaluation score of the target object and the scene score.

[0141] In this embodiment, the evaluation score of any initial autonomous driving image is calculated as follows: the evaluation scores of all target objects in the current initial autonomous driving image are accumulated to obtain an accumulated result, and then the accumulated result is multiplied by the scene score to obtain the evaluation score of the current initial autonomous driving image.

[0142] For example, the formula for determining the evaluation score of the initial autonomous driving image is as follows:

[0143]

[0144] Wherein, img_value_score represents the evaluation score of the initial autonomous driving image, scene_score represents the scene score, and bbox_value_score represents the evaluation score of the target object.

[0145] In S208, the target autonomous driving image dataset is filtered from the initial autonomous driving image dataset according to the evaluation scores of the plurality of initial autonomous driving images, and the labeling in the target autonomous driving image dataset is adjusted.

[0146] In this embodiment, the initial autonomous driving images in the initial autonomous driving image dataset can be sorted from high to low according to the evaluation scores of the initial autonomous driving images, a set number of initial autonomous driving images sorted in the front are extracted as a set number of target autonomous driving images, and the labeling corresponding to the set number of target autonomous driving images is adjusted.

[0147] Optionally, filtering the target autonomous driving image dataset from the initial autonomous driving image dataset according to the evaluation scores of the plurality of initial autonomous driving images comprises: sorting the plurality of initial autonomous driving images according to the evaluation scores of the plurality of initial autonomous driving images to obtain a sorted initial autonomous driving image dataset; and filtering a set number of target autonomous driving image datasets from the sorted initial autonomous driving image dataset.

[0148] Specifically, the plurality of initial autonomous driving images are sorted from high to low according to the evaluation scores of the plurality of initial autonomous driving images to obtain a sorted initial autonomous driving image dataset; and a set number of target autonomous driving image datasets sorted in the front are filtered from the sorted initial autonomous driving image dataset, that is, a set number of initial autonomous driving images in the front are extracted as a set number of target autonomous driving images. Alternatively, the plurality of initial autonomous driving images are sorted from low to high according to the evaluation scores of the plurality of initial autonomous driving images to obtain a sorted initial autonomous driving image dataset; and a set number of target autonomous driving image datasets sorted in the back are filtered from the sorted initial autonomous driving image dataset, that is, a set number of initial autonomous driving images in the back are extracted as a set number of target autonomous driving images.

[0149] In this embodiment, a set number of target autonomous driving image datasets are filtered from the sorted initial autonomous driving image dataset, so that the target autonomous driving images can be accurately filtered.

[0150] Exemplarily, Figure 8 An effect schematic diagram of a first image detection result provided by the embodiment of the application; Figure 9 An effect schematic diagram of a second image detection result provided by the embodiment of the application. Figure 8a first image detection result detected by a first automatic driving detection model, Figure 9 a second image detection result detected by a second automatic driving detection model. Figure 8 and Figure 9 As can be seen from the above, the first automatic driving detection model detects a group of bicycles with only one frame, while the second automatic driving detection model detects each bicycle separately, so that all the frames of the bicycles are not matched, and the comparison type is FN or FP; Figure 8 or Figure 9 The last column in the above, the first automatic driving detection model detects, the second automatic driving detection model does not detect, the comparison type is FN, and the value score is high; therefore, the evaluation score of the target object is mostly the evaluation score of FN and FP, and the score is very high.

[0151] Figure 10 A schematic structural diagram of an image data set processing device provided by an embodiment of the present disclosure is shown in FIG. 1. Figure 10 As shown in FIG. 1, the device comprises an initial automatic driving image data set acquisition module 1001, an image detection result obtaining module 1002, an image detection result comparison module 1003, a comparison type determination module 1004, an evaluation score determination module 1005, and a target automatic driving image data set screening module 1006.

[0152] The initial automatic driving image data set acquisition module 1001 is configured to acquire an initial automatic driving image data set; wherein the initial automatic driving image data set comprises a plurality of initial automatic driving images and corresponding labels.

[0153] The image detection result obtaining module 1002 is configured to input the initial image data set into a first automatic driving detection model and a second automatic driving detection model respectively, and obtain a first image detection result and a second image detection result respectively.

[0154] The image detection result comparison module 1003 is configured to compare the first image detection result and the second image detection result to obtain a comparison result.

[0155] The comparison type determination module 1004 is configured to determine a comparison type according to the comparison result; wherein the first image detection result is taken as a reference detection result; and the comparison type comprises true positive, false positive and false negative.

[0156] The evaluation score determination module 1005 is configured to determine an evaluation score of the plurality of initial automatic driving images according to the comparison type.

[0157] The target automatic driving image data set screening module 1006 is configured to screen a target automatic driving image data set from the initial automatic driving image data set according to the evaluation scores of the plurality of initial automatic driving images, and adjust the labels in the target automatic driving image data set.

[0158] The technical scheme disclosed in the embodiment comprises the following steps: obtaining an initial automatic driving image data set through an initial automatic driving image data set obtaining module; wherein the initial automatic driving image data set comprises a plurality of initial automatic driving images and corresponding labels; inputting the initial image data set into a first automatic driving detection model and a second automatic driving detection model respectively through an image detection result obtaining module to obtain a first image detection result and a second image detection result respectively; comparing the first image detection result and the second image detection result through an image detection result comparing module to obtain a comparison result; determining a comparison type according to the comparison result through a comparison type determining module; wherein the first image detection result is taken as a reference detection result; the comparison type comprises true positive, false positive and false negative; determining evaluation scores of the plurality of initial automatic driving images according to the comparison type through an evaluation score determining module; and screening a target automatic driving image data set from the initial automatic driving image data set according to the evaluation scores of the plurality of initial automatic driving images through a target automatic driving image data set screening module, and adjusting the labels in the target automatic driving image data set. According to the embodiment, the comparison type is determined according to the comparison result between the first image detection result and the second image detection result, the evaluation scores of the initial automatic driving images are determined according to the comparison type, and the target automatic driving image data set is screened according to the evaluation scores of the initial automatic driving images, so that the accuracy of image data set processing can be improved.

[0159] Optionally, each initial automatic driving image comprises at least one target object; the first image detection result comprises first bounding box information of the at least one target object, first categories corresponding to the at least one target object respectively, first confidence degrees of the first categories and first areas corresponding to the first bounding box information; the second image detection result comprises second bounding box information of the at least one target object, second categories corresponding to the at least one target object respectively, second confidence degrees corresponding to the second categories, and second areas corresponding to the second bounding box information; and the comparison result comprises a category comparison result. Optionally, the image detection result comparing module is specifically configured to compare the first categories and the second categories to obtain the category comparison result; wherein the category comparison result comprises category comparison consistency and category comparison inconsistency.

[0160] Optionally, the comparison type determination module is specifically configured to: determine an intersection over union according to the first bounding box information and the second bounding box information; determine a first intersection over union set value according to the first area and / or the second area; if the intersection over union is greater than or equal to the first intersection over union set value and the category comparison result is category comparison consistent, determine the comparison type as a true positive example; if the intersection over union is less than the first intersection over union set value and / or the category comparison result is category comparison inconsistent, determine the comparison type as a false positive example and a false negative example; or the first image detection result includes the first bounding box information of the target object, and the second image detection result does not include the second bounding box information of the target object, then determine the comparison type as a false negative example; the first image detection result does not include the first bounding box information of the target object, and the second image detection result includes the second bounding box information of the target object, then determine the comparison type as a false positive example.

[0161] Optionally, the evaluation score determination module is specifically configured to: for the evaluation score of any one initial automatic driving image, determine the evaluation score of the target object in the initial automatic driving image according to the comparison type; determine the scene score of the initial automatic driving image; and determine the evaluation score of the initial automatic driving image according to the evaluation score of the target object and the scene score.

[0162] Optionally, the evaluation score determination module is further configured to: determine at least one of an intersection over union score, a confidence score and an area score of the target object according to the comparison type; obtain a pre-set category score corresponding to the target object; and determine the evaluation score of the target object according to at least one of the intersection over union score, the confidence score, the area score and the category score.

[0163] Optionally, the evaluation score determination module is further configured to: determine a second intersection over union set value according to the first area and / or the second area; wherein the second intersection over union set value is greater than the first intersection over union set value; and determine an intersection over union evaluation score according to the comparison type, the first intersection over union set value, the second intersection over union set value and the intersection over union.

[0164] Optionally, the evaluation score determination module is further configured to: if the comparison type is false positive or false negative, and the intersection over union is less than the first intersection over union set value, the intersection over union evaluation score is a first set intersection over union evaluation score; if the comparison type is true positive, and the intersection over union is greater than or equal to the second intersection over union set value, the intersection over union evaluation score is a second set intersection over union evaluation score; if the comparison type is true positive, and the intersection over union falls within an interval formed by the first intersection over union set value and the second intersection over union set value, the first intersection over union evaluation score is determined according to the second intersection over union set value and the intersection over union, the second intersection over union evaluation score is determined according to the second intersection over union set value and the first intersection over union set value, and the intersection over union evaluation score is determined according to the first intersection over union evaluation score and the second intersection over union evaluation score.

[0165] Optionally, the evaluation score determination module is further configured to: if the comparison type is false positive, a first evaluation score of the second confidence is determined, and the first evaluation score of the second confidence is taken as the confidence score of the target object; if the comparison type is false negative, a first evaluation score of the first confidence is determined, and the first evaluation score of the first confidence is taken as the confidence score of the target object; if the comparison type is true positive, a second evaluation score of the first confidence and a second evaluation score of the second confidence are respectively determined, and the confidence score of the target object is determined according to the second evaluation score of the first confidence and the second evaluation score of the second confidence.

[0166] Optionally, the evaluation score determination module is further configured to: obtain a first confidence threshold and a second confidence threshold; wherein the second confidence threshold is greater than the first confidence threshold; if the first confidence or the second confidence is less than the first confidence threshold, take the first set confidence score as a first evaluation score of the first confidence or a first evaluation score of the second confidence; if the first confidence is greater than or equal to the second confidence threshold or the second confidence is greater than or equal to the second confidence threshold, take the first set confidence score as a first evaluation score of the first confidence or a first evaluation score of the second confidence; if the first confidence falls within an interval formed by the first confidence threshold and the second confidence threshold, determine a first confidence score according to the first confidence, the second confidence threshold and the first confidence threshold; determine a second confidence score according to the second confidence threshold and the first confidence threshold; determine a first evaluation score of the first confidence according to the first confidence score and the second confidence score; or if the second confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, determine a third confidence score according to the second confidence, the second confidence threshold and the first confidence threshold; determine a fourth confidence score according to the second confidence threshold and the first confidence threshold; determine a first evaluation score of the second confidence according to the third confidence score and the fourth confidence score.

[0167] Optionally, the evaluation score determination module is further configured to: if the first confidence or the second confidence is less than the first confidence threshold, take the first set confidence score as a second evaluation score of the first confidence or a second evaluation score of the second confidence; if the first confidence is greater than or equal to the first confidence threshold or the second confidence is greater than or equal to the second confidence threshold, take the first set confidence score as a second evaluation score of the first confidence or a second evaluation score of the second confidence; if the first confidence falls within an interval formed by the first confidence threshold and the second confidence threshold, determine a fifth confidence score according to the first confidence and the second confidence threshold; determine a sixth confidence score according to the second confidence threshold and the first confidence threshold; determine a second evaluation score of the first confidence according to the fifth confidence score and the sixth confidence score; or if the second confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, determine a seventh confidence score according to the second confidence and the second confidence threshold; determine an eighth confidence score according to the second confidence threshold and the first confidence threshold; determine a second evaluation score of the second confidence according to the seventh confidence score and the eighth confidence score.

[0168] Optionally, the evaluation score determining module is further configured to: if the comparison type is false positive, determine a first evaluation score of the second area, and take the first evaluation score of the second area as the area score of the target object; if the comparison type is false negative, determine a first evaluation score of the first area, and take the first evaluation score of the first area as the area score of the target object; and if the comparison type is true positive, determine a second evaluation score of the first area and a second evaluation score of the second area respectively, and determine the area score of the target object according to the second evaluation score of the first area and the second evaluation score of the second area.

[0169] Optionally, the evaluation score determining module is further configured to: obtain a first area threshold and a second area threshold, wherein the second area threshold is greater than the first area threshold; if the first area or the second area is less than the first area threshold, take a first set area score as the first evaluation score of the first area or the first evaluation score of the second area; if the first area is greater than or equal to the second area threshold, or the second area is greater than or equal to the second area threshold, take a second set area score as the first evaluation score of the first area or the first evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and the second area threshold, determine a first area score according to the first area, the second area threshold and the first area threshold; determine a second area score according to the second area threshold and the first area threshold; and determine the first evaluation score of the first area according to the first area score and the second area score; or if the second area falls within the interval formed by the first area threshold and the second area threshold, determine a third area score according to the second area, the second area threshold and the first area threshold; determine a fourth area score according to the second area threshold and the first area threshold; and determine the first evaluation score of the second area according to the third area score and the fourth area score.

[0170] Optionally, the evaluation score determination module is further configured to: if the first area or the second area is less than a first area threshold, determine a first set area score as the second evaluation score of the first area or the second evaluation score of the second area; if the first area is greater than or equal to the first area threshold, or if the second area is greater than or equal to the first area threshold, determine a second set area score as the second evaluation score of the first area or the second evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and the second area value, determine a fifth area score according to the first area and the first area threshold; determine a sixth area score according to the second area threshold and the first area threshold; determine the second evaluation score of the first area according to the fifth area score and the sixth area score; or if the second area falls within the interval formed by the first area threshold and the second area value, determine a seventh area score according to the second area and the first area threshold; determine an eighth area score according to the second area threshold and the first area threshold; and determine the second evaluation score of the second area according to the seventh area score and the eighth area score.

[0171] Optionally, the evaluation score determination module is further configured to: input the initial automatic driving image into an automatic driving scene model to output predicted scene information; obtain set scene information corresponding to the initial automatic driving image; determine a scene similarity between the predicted scene information and the set scene information; and determine a scene score of the initial automatic driving image according to the scene similarity.

[0172] Optionally, the target automatic driving image data set screening module is specifically configured to: sort the plurality of initial automatic driving images according to the evaluation scores of the plurality of initial automatic driving images to obtain a sorted initial automatic driving image data set; and screen a set number of target automatic driving image data sets from the sorted initial automatic driving image data set.

[0173] The image data set processing apparatus provided in the embodiments of the present disclosure can execute the image data set processing method provided in any of the embodiments of the present disclosure, and has the corresponding function modules and beneficial effects of the execution method.

[0174] It should be noted that each unit and module included in the above apparatus is only divided according to the function logic, but is not limited to the above division, as long as the corresponding function can be implemented; in addition, the specific name of each functional unit is only for convenient mutual distinction, and does not serve to limit the protection scope of the embodiments of the present disclosure.

[0175] Figure 11 A structural schematic diagram of an electronic device provided by the embodiments of the present disclosure is shown in FIG. 1. The following refers to the structural schematic diagram of the electronic device shown in FIG. 1. Figure 11It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 11 The diagram below shows the structure of the terminal device or server 1100. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 11 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0176] like Figure 11 As shown, electronic device 1100 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 1101, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1102 or a program loaded from storage device 1108 into random access memory (RAM) 1103. The RAM 1103 also stores various programs and data required for the operation of electronic device 1100. The processing unit 1101, ROM 1102, and RAM 1103 are interconnected via bus 1104. An edit / output (I / O) interface 1105 is also connected to bus 1104.

[0177] Typically, the following devices can be connected to I / O interface 1105: input devices 1106 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 1107 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1108 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1109. Communication device 1109 allows electronic device 1100 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 11 An electronic device 1100 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0178] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for executing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network by the communication apparatus 1109, or installed from the storage apparatus 1108, or installed from the ROM 1102. When the computer program is executed by the processing apparatus 1101, the above-mentioned functions defined in the methods of embodiments of the present disclosure are executed.

[0179] The names of the messages or information exchanged between the plurality of devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of the messages or information.

[0180] The electronic device provided by the embodiments of the present disclosure and the image dataset processing method provided by the above-mentioned embodiments belong to the same inventive concept, and the technical details not described in detail in the present embodiment can be referred to the above-mentioned embodiments, and the present embodiment has the same beneficial effects as the above-mentioned embodiments.

[0181] The embodiments of the present disclosure provide a computer storage medium, which stores a computer program, and the program is executed by a processor to implement the image dataset processing method provided by the above-mentioned embodiments.

[0182] It should be noted that the computer-readable medium described above can be a computer-readable signal medium or a computer-readable storage medium or any combination thereof. The computer-readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any suitable combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program used by or in connection with an instruction execution system, apparatus or device. In the present disclosure, the computer-readable signal medium can include a data signal propagated in baseband or propagated as a carrier wave in a propagated data signal, in which the computer-readable program code is contained. Such a propagated data signal can take many forms, including but not limited to, an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium that can send, propagate or transfer the program for use by or in connection with the instruction execution system, apparatus or device. The program code contained in the computer-readable medium can be transmitted by any suitable medium, including but not limited to, wire, cable, RF (radio frequency), etc., or any suitable combination of the above.

[0183] In some embodiments, the client, server, or both can communicate using any current known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and can be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include local area networks ("LAN"), wide area networks ("WAN"), the Internet, and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any current known or future developed networks.

[0184] The computer-readable medium described above can be included in the electronic device described above; or can exist separately from the electronic device, and not be assembled into the electronic device.

[0185] The computer readable medium described above carries one or more programs, when the one or more programs are executed by the electronic device, cause the electronic device to: obtain an initial autonomous driving image data set; wherein the initial autonomous driving image data set includes a plurality of initial autonomous driving images and corresponding labels; input the initial image data set into a first autonomous driving detection model and a second autonomous driving detection model respectively, and obtain a first image detection result and a second image detection result respectively; compare the first image detection result and the second image detection result to obtain a comparison result; determine a comparison type according to the comparison result; wherein the first image detection result is taken as a reference detection result; the comparison type includes true positive, false positive and false negative; determine evaluation scores of the plurality of initial autonomous driving images according to the comparison type; and screen a target autonomous driving image data set from the initial autonomous driving image data set according to the evaluation scores of the plurality of initial autonomous driving images, and adjust the labels in the target autonomous driving image data set.

[0186] Computer program code for carrying out operations of the present disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0187] The computer program product of the first aspect can include one or more non-transitory computer-readable media storing instructions that, when executed, cause one or more processors to perform the operations of the first aspect. The one or more non-transitory computer-readable media can include, for example, magnetic media such as one or more magnetic disks, magnetic tapes or cassettes; optical media such as one or more compact discs, optical discs or Blu-ray discs; magneto-optical media such as one or more floptical discs; solid state media such as one or more solid state drives or other flash memory arrays; or any suitable combination of these. The one or more non-transitory computer-readable media can be encoded with instructions that, when executed, cause one or more processors to perform the operations of the first aspect.

[0188] The units described in the embodiments of the present disclosure can be implemented by software, or by hardware, or by a combination of software and hardware. In some cases, the name of the unit does not constitute a limitation on the unit itself. For example, the first obtaining unit can also be described as a unit for obtaining at least two Internet protocol addresses.

[0189] The functions described above in the detailed description of the above embodiments can be performed by one or more hardware logic components. For example, and without limitation, illustrative types of hardware logic components that can be used include Field-programmable Gate Arrays (FPGAs), Program-specific Integrated Circuits (ASICs), Program-specific Standard Products (ASSPs), System-on-a-chip systems (SOCs), Complex Programmable Logic Devices (CPLDs), etc.

[0190] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more of: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0191] The above description merely illustrates the preferred embodiments of the disclosure and a principle for applying the technologies. It is understood by those skilled in the art that the disclosed scope of the disclosure is not limited to the technical solutions formed by the specific combinations of the technical features described above, and should also cover other technical solutions formed by the combinations of the technical features described above or their equivalent features without departing from the disclosed concept. For example, the technical solutions formed by the mutual replacement of the above-described features and the technical features with similar functions disclosed in the disclosure (but not limited to) can be used.

[0192] Further, although operations are depicted in a particular, sequential order, this should not be understood as requiring or implying that the operations are performed in the order illustrated or sequentially. In certain circumstances, multitasking and parallel processing can be advantageous. Likewise, although specific implementation details are included for the purpose of providing a thorough disclosure, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub-combination.

[0193] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A method of processing image data sets, characterized by The method comprises the following steps: obtaining an initial automatic driving image dataset; wherein the initial automatic driving image dataset comprises a plurality of initial automatic driving images and corresponding labels; inputting the initial automatic driving image dataset into a first automatic driving detection model and a second automatic driving detection model respectively to obtain first image detection results and second image detection results respectively; comparing the first image detection results and the second image detection results to obtain comparison results; determining a comparison type according to the comparison results; wherein the first image detection results are taken as reference detection results; the comparison type comprises true positive, false positive and false negative; determining evaluation scores of the plurality of initial automatic driving images according to the comparison type; screening a target automatic driving image dataset from the initial automatic driving image dataset according to the evaluation scores of the plurality of initial automatic driving images, and adjusting the labels in the target automatic driving image dataset; each initial automatic driving image comprises at least one target object; the first image detection results comprise first bounding box information of the at least one target object, first categories corresponding to the at least one target object respectively, first confidence degrees of the first categories and first areas corresponding to the first bounding box information; the second image detection results comprise second bounding box information of the at least one target object, second categories corresponding to the at least one target object respectively, second confidence degrees corresponding to the second categories and second areas corresponding to the second bounding box information; the comparison results comprise category comparison results; comparing the first categories and the second categories to obtain the category comparison results; wherein the category comparison results comprise category comparison consistency and category comparison inconsistency; determining a comparison type according to the comparison results, comprising: determining an intersection over union according to the first bounding box information and the second bounding box information; determining a first intersection over union setting value according to the first area and / or the second area; if the intersection over union is greater than or equal to the first intersection over union setting value and the category comparison result is the category comparison consistency, the comparison type is determined as true positive; if the intersection over union is less than the first intersection over union setting value and / or the category comparison result is the category comparison inconsistency, the comparison type is determined as false positive and false negative; or; if the first image detection results comprise first bounding box information of a target object and the second image detection results do not comprise second bounding box information of the target object, the comparison type is determined as false negative; if the first image detection results do not comprise first bounding box information of a target object and the second image detection results comprise second bounding box information of the target object, the comparison type is determined as false positive. determining evaluation scores of the plurality of initial automatic driving images according to the comparison type, comprising:

2. The method of claim 1, wherein, for the evaluation score of any one initial automatic driving image, determining evaluation scores of target objects in the initial automatic driving image according to the comparison type. ​ determine a scene score of the initial autonomous driving image; determine an evaluation score of the initial autonomous driving image according to the evaluation score of the target object and the scene score.

3. The method according to claim 2, characterized in that, determine an evaluation score of a target object in an initial autonomous driving image according to a comparison type, comprising: determine at least one of an intersection over union score, a confidence score and an area score of the target object according to the comparison type; obtain a pre-set category score corresponding to the target object; determine the evaluation score of the target object according to at least one of the intersection over union score, the confidence score, the area score and the category score.

4. The method of claim 3, wherein, determine an intersection over union score of a target object according to a comparison type, comprising: determine a second intersection over union setting value according to the first area and / or the second area; wherein the second intersection over union setting value is greater than the first intersection over union setting value; determine an intersection over union evaluation score according to the comparison type, the first intersection over union setting value, the second intersection over union setting value and the intersection over union.

5. The method of claim 4, wherein, determine an intersection over union evaluation score according to the comparison type, the first intersection over union setting value, the second intersection over union setting value and the intersection over union, comprising: if the comparison type is a false positive or a false negative, and the intersection over union is less than the first intersection over union setting value, then the intersection over union evaluation score is a first set intersection over union evaluation score; if the comparison type is a true positive, and the intersection over union is greater than or equal to the second intersection over union setting value, then the intersection over union evaluation score is a second set intersection over union evaluation score; if the comparison type is a true positive, and the intersection over union falls within an interval formed by the first intersection over union setting value and the second intersection over union setting value, then determine a first intersection over union evaluation score according to the second intersection over union setting value and the intersection over union; determine a second intersection over union evaluation score according to the second intersection over union setting value and the first intersection over union setting value; determine the intersection over union evaluation score according to the first intersection over union evaluation score and the second intersection over union evaluation score.

6. The method of claim 3, wherein, determine a confidence score of a target object according to a comparison type, comprising: if the comparison type is a false positive, then determine a first evaluation score of a second confidence, and take the first evaluation score of the second confidence as the confidence score of the target object; if the comparison type is a false negative, then determine a first evaluation score of a first confidence, and take the first evaluation score of the first confidence as the confidence score of the target object; if the comparison type is a true positive, then determine a second evaluation score of the first confidence and a second evaluation score of the second confidence respectively, and determine the confidence score of the target object according to the second evaluation score of the first confidence and the second evaluation score of the second confidence.

7. The method of claim 6, wherein, determine the first evaluation score of the first confidence, or determine the first evaluation score of the second confidence, comprising: obtain a first confidence threshold and a second confidence threshold; wherein the second confidence threshold is greater than the first confidence threshold; if the first confidence or the second confidence is less than the first confidence threshold, then take a first set confidence score as the first evaluation score of the first confidence or the first evaluation score of the second confidence; if the first confidence is greater than or equal to a second confidence threshold or the second confidence is greater than or equal to the second confidence threshold, then determining a second set confidence score as the first evaluation score of the first confidence or the first evaluation score of the second confidence; if the first confidence falls within an interval formed by the first confidence threshold and the second confidence threshold, then determining a first confidence score according to the first confidence, the second confidence threshold and the first confidence threshold; determining a second confidence score according to the second confidence threshold and the first confidence threshold; determining the first evaluation score of the first confidence according to the first confidence score and the second confidence score; or; if the second confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, then determining a third confidence score according to the second confidence, the second confidence threshold and the first confidence threshold; determining a fourth confidence score according to the second confidence threshold and the first confidence threshold; determining the first evaluation score of the second confidence according to the third confidence score and the fourth confidence score.

8. The method of claim 7, wherein, determining a second evaluation score of the first confidence or determining a second evaluation score of the second confidence, comprising: if the first confidence or the second confidence is less than a first confidence threshold, then determining a first set confidence score as the second evaluation score of the first confidence or the second evaluation score of the second confidence; if the first confidence is greater than or equal to the first confidence threshold or the second confidence is greater than or equal to the second confidence threshold, then determining a first set confidence score as the second evaluation score of the first confidence or the second evaluation score of the second confidence; if the first confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, then determining a fifth confidence score according to the first confidence and the second confidence threshold; determining a sixth confidence score according to the second confidence threshold and the first confidence threshold; determining the second evaluation score of the first confidence according to the fifth confidence score and the sixth confidence score; or; if the second confidence falls within the interval formed by the first confidence threshold and the second confidence threshold, then determining a seventh confidence score according to the second confidence and the second confidence threshold; determining an eighth confidence score according to the second confidence threshold and the first confidence threshold; determining the second evaluation score of the second confidence according to the seventh confidence score and the eighth confidence score.

9. The method of claim 3, wherein, determining an area score of the target object according to the comparison type, comprising: if the comparison type is false positive, then determining a first evaluation score of a second area, and determining the first evaluation score of the second area as the area score of the target object; if the comparison type is false negative, then determining a first evaluation score of a first area, and determining the first evaluation score of the first area as the area score of the target object; If the comparison type is a true case, a second evaluation score of the first area and a second evaluation score of the second area are determined respectively, and an area score of the target object is determined according to the second evaluation score of the first area and the second evaluation score of the second area.

10. The method of claim 9, wherein, The first evaluation score of the first area is determined, or the first evaluation score of the second area is determined, including: obtaining a first area threshold and a second area threshold; wherein the second area threshold is greater than the first area threshold; if the first area or the second area is less than the first area threshold, a first set area score is taken as the first evaluation score of the first area or the first evaluation score of the second area; if the first area is greater than or equal to the second area threshold, or if the second area is greater than or equal to the second area threshold, a second set area score is taken as the first evaluation score of the first area or the first evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and the second area threshold, a first area score is determined according to the first area, the second area threshold and the first area threshold; a second area score is determined according to the second area threshold and the first area threshold; a first evaluation score of the first area is determined according to the first area score and the second area score; or; if the second area falls within the interval formed by the first area threshold and the second area threshold, a third area score is determined according to the second area, the second area threshold and the first area threshold; a fourth area score is determined according to the second area threshold and the first area threshold; a first evaluation score of the second area is determined according to the third area score and the fourth area score.

11. The method of claim 10, wherein, The second evaluation score of the first area is determined, or the second evaluation score of the second area is determined, including: if the first area or the second area is less than the first area threshold, a first set area score is taken as the second evaluation score of the first area or the second evaluation score of the second area; if the first area is greater than or equal to the first area threshold, or if the second area is greater than or equal to the first area threshold, a second set area score is taken as the second evaluation score of the first area or the second evaluation score of the second area; if the first area falls within an interval formed by the first area threshold and the second area value, a fifth area score is determined according to the first area and the first area threshold; a sixth area score is determined according to the second area threshold and the first area threshold; a second evaluation score of the first area is determined according to the fifth area score and the sixth area score; or; if the second area falls within the interval formed by the first area threshold and the second area value, a seventh area score is determined according to the second area and the first area threshold; an eighth area score is determined according to the second area threshold and the first area threshold; a second evaluation score of the second area is determined according to the seventh area score and the eighth area score.

12. The method of claim 2, wherein, The scene score of the initial automatic driving image is determined, including: inputting the initial automatic driving image into an automatic driving scene model to output predicted scene information; obtain scene information corresponding to the initial automatic driving image; determine scene similarity between the predicted scene information and the set scene information; determine a scene score of the initial automatic driving image according to the scene similarity.

13. The method of claim 1, wherein, According to the evaluation score of the plurality of initial automatic driving images, a target automatic driving image dataset is selected from the initial automatic driving image dataset, comprising: According to the evaluation score of the plurality of initial automatic driving images, the plurality of initial automatic driving images are sorted to obtain a sorted initial automatic driving image dataset; A target automatic driving image dataset is selected from the sorted initial automatic driving image dataset.

14. An image data set processing apparatus, characterized by Comprising: An initial automatic driving image dataset acquisition module is configured to acquire an initial automatic driving image dataset; wherein the initial automatic driving image dataset comprises a plurality of initial automatic driving images and corresponding labels; An image detection result obtaining module is configured to input the initial automatic driving image dataset into a first automatic driving detection model and a second automatic driving detection model respectively, and obtain a first image detection result and a second image detection result respectively; An image detection result comparison module is configured to compare the first image detection result and the second image detection result to obtain a comparison result; A comparison type determination module is configured to determine a comparison type according to the comparison result; wherein the first image detection result is used as a reference detection result; the comparison type includes true positive, false positive and false negative; An evaluation score determination module is configured to determine the evaluation score of the plurality of initial automatic driving images according to the comparison type; A target automatic driving image dataset selection module is configured to select a target automatic driving image dataset from the initial automatic driving image dataset according to the evaluation score of the plurality of initial automatic driving images, and adjust the labels in the target automatic driving image dataset; Each initial automatic driving image comprises at least one target object; the first image detection result comprises first bounding box information of the at least one target object, first categories corresponding to the at least one target object respectively, first confidence of the first categories, and first areas corresponding to the first bounding box information; the second image detection result comprises second bounding box information of the at least one target object, second categories corresponding to the at least one target object respectively, second confidence corresponding to the second categories, and second areas corresponding to the second bounding box information; the comparison result comprises a category comparison result, and the image detection result comparison module is specifically configured to: compare the first categories and the second categories to obtain a category comparison result; wherein the category comparison result includes category comparison consistency and category comparison inconsistency; The comparison type determination module is specifically configured to: determine an intersection-over-union ratio according to the first bounding box information and the second bounding box information; determine a first intersection-over-union ratio set value according to the first area and / or the second area; if the intersection-over-union ratio is greater than or equal to the first intersection-over-union ratio set value and the category comparison result is category comparison consistent, determine the comparison type as a true example; if the intersection-over-union ratio is less than the first intersection-over-union ratio set value and / or the category comparison result is category comparison inconsistent, determine the comparison type as a false positive example and a false negative example; or; the first image detection result includes first bounding box information of a target object, and the second image detection result does not include second bounding box information of the target object, then determine the comparison type as a false negative example; the first image detection result does not include first bounding box information of a target object, and the second image detection result includes second bounding box information of the target object, then determine the comparison type as a false positive example.

15. An electronic device, comprising: The electronic device includes: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the image data set processing method as claimed in any one of claims 1-13.

16. A storage medium containing computer-executable instructions for performing the image data set processing method as claimed in any one of claims 1-13 when executed by a computer processor.

Images