Image set generation, sample set generation, automated testing methods and apparatus

Abstract

The disclosure provides an image set generation method and device, relates to the technical field of artificial intelligence, in particular to the technical field of computer vision, deep learning and the like. The specific implementation scheme is: obtaining a current target image based on current visual key information; sending the target image to an intelligent recognition interaction device to obtain recognition requirement information of a corresponding next image output by the intelligent recognition interaction device; in response to a current target image set including the target image not meeting the requirements of the recognition requirement information, obtaining updated visual key information based on the recognition requirement information; obtaining a next image based on the updated visual key information, and updating the target image set through the next image. The embodiment improves the generation efficiency of visual related resources.

Description

Technical Field

[0001] This disclosure relates to the field of computer application technology, the field of artificial intelligence technology, specifically to the fields of computer vision, deep learning and other technical fields, and in particular to an image set generation method and apparatus, a sample set generation method and apparatus, an automated testing method and apparatus, an electronic device, a computer-readable medium and a computer program product. Background Technology

[0002] Traditional visual testing requires video or image resources that are manually searched and stored locally or in corresponding project documents. This makes sharing difficult, maintenance costs high, and is not conducive to implementing complex testing strategies or reusing across multiple platforms. Summary of the Invention

[0003] An image set generation method and apparatus, a sample set generation method and apparatus, an automated testing method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product are provided.

[0004] According to a first aspect, an image set generation method is provided, the method comprising: obtaining a current target image based on current visual key information; sending the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device; in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, obtaining updated visual key information based on the recognition requirement information; obtaining the next image based on the updated visual key information, and updating the target image set with the next image.

[0005] According to the second aspect, a sample set generation method is provided, the method comprising: obtaining a current target image based on current visual key information; sending the target image to an intelligent recognition interaction device to obtain image annotation information of the target image and recognition requirement information of the corresponding next image output by the intelligent recognition interaction device; associating the target image and the image annotation information to obtain image samples; in response to the current image sample set including image samples not meeting the requirements of the recognition requirement information, obtaining updated visual key information based on the recognition requirement information; obtaining the next image sample based on the updated visual key information and the intelligent recognition interaction device, and updating the image sample set with the next image sample.

[0006] According to the third aspect, an automated testing method is provided, which includes: selecting image samples related to the action to be tested from the current image sample set to obtain selected samples; obtaining the image sample set based on the sample set generation method of any implementation of the second aspect; and performing automated testing on the intelligent recognition interactive device based on the selected samples to obtain the automated test results of the intelligent recognition interactive device.

[0007] According to a fourth aspect, an image set generation apparatus is provided, the apparatus comprising: a target image obtaining unit configured to obtain a current target image based on current visual key information; a target information obtaining unit configured to send the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device; an updating unit configured to, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, obtain updated visual key information based on the recognition requirement information; and a target set obtaining unit configured to obtain the next image based on the updated visual key information and update the target image set using the next image.

[0008] According to a fifth aspect, a sample set generation apparatus is provided, comprising: a sample image obtaining unit configured to obtain a current target image based on current visual key information; a recognition unit configured to send the target image to an intelligent recognition interaction device to obtain image annotation information of the target image output by the intelligent recognition interaction device and recognition requirement information of the corresponding next image; an association unit configured to associate the target image and the image annotation information to obtain image samples; a sample information obtaining unit configured to obtain updated visual key information based on the recognition requirement information in response to the current image sample set including image samples not meeting the requirements of the recognition requirement information; and a sample set obtaining unit configured to obtain the next image sample based on the updated visual key information and the intelligent recognition interaction device, and update the image sample set with the next image sample.

[0009] According to the sixth aspect, an automated testing apparatus is provided, the apparatus comprising: a selection unit configured to select image samples related to the action to be tested from a current image sample set to obtain selected samples; the image sample set is obtained based on a sample set generation apparatus according to any implementation of the fifth aspect; and a testing unit configured to perform automated testing on an intelligent recognition interactive device based on the selected samples to obtain automated test results of the intelligent recognition interactive device.

[0010] According to a seventh aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a method as described in any implementation of the first to third aspects.

[0011] According to the eighth aspect, a non-transitory computer-readable storage medium is provided that stores computer instructions for causing a computer to perform the method described in any implementation of the first to third aspects.

[0012] According to a ninth aspect, a computer program product is provided, including a computer program that, when executed by a processor, implements the method as described in any of the implementations of the first to third aspects.

[0013] The image set generation method and apparatus provided in the embodiments of this disclosure first obtain a current target image based on current visual key information; second, send the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device; third, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, obtain updated visual key information based on the recognition requirement information; finally, obtain the next image based on the updated visual key information, and update the target image set with the next image. Thus, in the process of obtaining the target image through visual key information, the visual key information is updated in real time through the recognition requirement information output by the intelligent recognition interaction device, allowing the target images in the target image set to gradually increase according to the device's needs, automatically expanding the target image set.

[0014] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0015] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0016] Figure 1 This is a flowchart of an embodiment of the image set generation method according to the present disclosure;

[0017] Figure 2 This is a flowchart of an embodiment of the sample set generation method according to the present disclosure;

[0018] Figure 3 This is a flowchart of an embodiment of the automated testing method according to the present disclosure;

[0019] Figure 4 This is a schematic diagram of the structure of an embodiment of the image set generation apparatus according to the present disclosure;

[0020] Figure 5 This is a schematic diagram of the structure of an embodiment of the sample set generation apparatus according to the present disclosure;

[0021] Figure 6 This is a schematic diagram of the structure of an embodiment of the automated testing apparatus according to the present disclosure;

[0022] Figure 7This is a block diagram of an electronic device used to implement the image set generation method, sample set generation method, or automated testing method of the embodiments of this disclosure. Detailed Implementation

[0023] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0024] In this embodiment, "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" and "second" may explicitly or implicitly include at least one of those features.

[0025] This disclosure provides a method for generating image sets. Figure 1 A flow 100 of an embodiment of the image set generation method according to the present disclosure is shown, the image set generation method including the following steps:

[0026] Step 101: Obtain the current target image based on the current key visual information.

[0027] In this embodiment, visual key information is text data related to the target image to be obtained. Visual key information may include: image format, image type, image label, and attribute information of the target in the image. The target is a scene, plant, animal, or other similar object presented in the target image. Target attribute information may include the target's action, the phenomenon it represents, and the target's posture. Visual key information describes the target image in the form of text data, and the target image can be generated based on this visual key information.

[0028] In this embodiment, the target image is an image containing a target within the image information. For example, if an image contains a child squatting, then the image is a target image where the target is the child. To obtain this target image, the key visual information could be: the image format is jpg, the target is a person, and the target's action is squatting.

[0029] In this embodiment, the current visual key information can be the initial visual key information directly input by the operator, or it can be the information obtained after extracting key information from the initial recognition requirement information of the intelligent recognition interaction device. For example, if the initial recognition requirement information output by the intelligent recognition interaction device is "recognize an image of a person standing," then the current visual key information is: person and person standing. Optionally, the current visual key information can also be the recognition requirement information obtained by the intelligent recognition interaction device after performing multiple rounds of image recognition, using the last recognition requirement information.

[0030] Step 102: Send the target image to the intelligent recognition interaction device to obtain the recognition requirement information for the next image output by the intelligent recognition interaction device.

[0031] In this embodiment, the intelligent recognition interactive device is a device that automatically identifies the target type or target attribute type of multiple targets in a target image. The device can provide recognition results for the target type or target attribute type of different targets, and the device can also automatically plan the target type or target attribute type of the target in the next image. The target type or target attribute type automatically planned by the device can be determined by the sorting model in the device (random or output according to the sequence requirements). For example, after the intelligent recognition interactive device recognizes a child in the current target image, it outputs the recognition requirement information for the next image, "Please enter an adult image"; or, after the intelligent recognition interactive device recognizes that the action type of the target in the current target image is squatting, it outputs the recognition requirement information for the next image, "Please enter a standing person image".

[0032] In this embodiment, the identification requirement information is text data output by the intelligent identification interaction device that is related to the type of target or the type of target attribute in the image. This text data can be obtained directly from the intelligent identification interaction device or by converting the speech of the intelligent identification interaction device. By identifying the requirement information, the different target types or target attribute types required by the intelligent identification interaction device can be determined.

[0033] In this embodiment, the output order of the identification requirement information can be obtained based on the requirements of the order sorting module set inside the intelligent identification interactive device. For example, if the intelligent identification interactive device includes a random sorting module, then after obtaining the current target image, the intelligent identification interactive device will randomly determine the target type or target attribute type of the next image.

[0034] In one example, the intelligent recognition interaction device is a device for recognizing a person's body movements. When a target image including a lying person is input into the intelligent recognition interaction device, the intelligent recognition interaction device recognizes the target image, obtains the recognition result of the target image, and outputs the recognition requirement information "Please input the next standing person image".

[0035] Step 103: In response to the current set of target images, including the target image, not meeting the requirements of the recognition requirement information, updated visual key information is obtained based on the recognition requirement information.

[0036] In this embodiment, the current target image set includes at least one target image. This at least one target image can be obtained through multiple visual key information sequences, each with different visual key information. Furthermore, each target image within the at least one target image set is also different. When all target images in the current target image set differ from the target type or target attribute type corresponding to the recognition requirement information, it is determined that the current target image set does not meet the requirements of the recognition requirement information.

[0037] In this embodiment, the identification requirement information includes: the type of target or the type of target attribute. The current target image set is a collection obtained by multiple target images. If the current target image set does not meet the requirements of the identification requirement information, and it is determined that the target type or the type of target attribute of the identification requirement information does not appear in the current target image set, then it is necessary to modify or replace the visual key information to obtain the updated visual key information. The latest target image can be obtained through the updated visual key information, so that the latest target image can be input into the intelligent recognition interaction device.

[0038] Step 104: Based on the updated visual key information, obtain the next image and update the target image set using the next image.

[0039] In this embodiment, after obtaining the updated visual key information, the next image can be obtained by searching the Internet using the updated visual key information.

[0040] Optionally, if the next image cannot be found on the internet, it can be generated based on the updated visual key information. It should be noted that the updated visual key information contains crucial information describing the next image; this crucial information, along with traditional mapping tools, can be used to generate the next image.

[0041] In this embodiment, after obtaining the next image, the next image is added as a target image to the current target image set to update the current target image set. As the visual key information is updated and iterated, the target images added to the target image set become larger and larger, gradually and automatically expanding the data volume of the target image set.

[0042] The image set generation method provided in this disclosure first obtains the current target image based on the current visual key information; second, it sends the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device; third, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, it obtains updated visual key information based on the recognition requirement information; finally, it obtains the next image based on the updated visual key information and updates the target image set with the next image. Therefore, in the process of obtaining the target image through visual key information, the visual key information is updated in real time through the recognition requirement information output by the intelligent recognition interaction device, allowing the target images in the target image set to gradually increase according to the device's needs, automatically expanding the target image set.

[0043] In this embodiment, the identification requirement information includes: image type, image format, and tag; after sending the target image to the intelligent recognition interaction device, in some embodiments of this disclosure, the above-mentioned image set generation method further includes: obtaining the target tag of the target image output by the intelligent recognition interaction device, tagging the target image based on the target tag, and adding the tagged target image to the current target image set, wherein each target image in the current target image set is an image with a tag.

[0044] In some optional implementations of this embodiment, the above-mentioned response to the current target image set including the target image not meeting the requirements of the recognition requirement information, and obtaining the updated visual key information based on the recognition requirement information, includes: searching for a target image corresponding to the recognition requirement information from the current target image set; determining that the target image set does not meet the requirements of the recognition requirement information in response to the absence of a target image corresponding to the recognition requirement information in the image sample set; and updating the visual key information based on the recognition requirement information to obtain the updated visual key information.

[0045] In this embodiment, after obtaining the target image based on the current visual key information, the target image is added to the target image set, so that the current target image set includes the target image. The target images in the target image set can be labeled with different tags for image type, image format, and target type or target attribute type. It should be noted that the target image set not only contains the current target image but also historical target images; the historical target images are the next images obtained through updated visual key information.

[0046] In this embodiment, the identification requirement information can be information with an information content equal to that of the visual key information. Both the identification requirement information and the visual key information include: target, target type, or target attribute type. For example, the identification requirement information may be: the target is a person and the target attribute type is a person crouching, while the visual key information may be: the target is an animal and the target attribute type is an animal running. Based on the identification requirement information, updating the visual key information may include: directly replacing the visual key information with the identification requirement information to obtain the updated visual key information.

[0047] The method for obtaining updated visual key information provided in the embodiments of this disclosure updates the visual key information based on the identification requirement information when there is no target image corresponding to the visual key information in the current target image set. The updated visual key information is then used as the current visual key information to determine the next image, providing a reliable way to effectively update the target image and ensuring the reliability of the visual key information update.

[0048] Optionally, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, the updated visual key information based on the recognition requirement information includes: determining the image type of the target image from the recognition requirement information; in response to the absence of a target image corresponding to the image type in the image sample set, determining that the current target image set does not meet the requirements of the recognition requirement, updating the visual key information based on the image type, so as to search for or generate a target image corresponding to the image type through the updated visual key information.

[0049] Optionally, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, obtaining the updated visual key information based on the recognition requirement information further includes: determining the image format of the target image from the recognition requirement information; in response to the absence of a target image corresponding to the image format in the image sample set, determining that the current target image set does not meet the requirements of the recognition requirement, updating the visual key information based on the image format, so as to search for or generate a target image corresponding to the image format through the updated visual key information.

[0050] In some optional implementations of this embodiment, obtaining the next image based on the updated visual key information and updating the target image set with the next image includes: searching for or generating the next image based on the updated visual key information; adding the next image to the target image set to obtain the updated target image set; and using the updated target image set as the current target image set.

[0051] In this optional implementation, the updated visual key information is the text data of the obtained image. The next image can be found by searching for the image on an internet search engine using this text data. If the next image cannot be found, it can be generated by an image generation tool.

[0052] The method for updating the target image set provided by this optional implementation searches for or generates the next image based on the updated visual key information, providing a reliable means for expanding the target image set and ensuring the effectiveness of the target image set expansion.

[0053] Optionally, obtaining the next image based on the updated visual key information and updating the target image set with the next image includes: searching for the next image based on the updated visual key information; if no next image is found, generating the next image based on the updated visual key information; adding the next image to the target image set to obtain the updated target image set; and using the updated target image set as the current target image set.

[0054] Further reference Figure 2 This disclosure provides a method for generating a sample set. Figure 2 A flow 200 is shown as an embodiment of the sample set generation method according to the present disclosure, the sample set generation method comprising the following steps:

[0055] Step 201: Obtain the current target image based on the current key visual information.

[0056] It should be understood that the operations and features in step 201 correspond to the operations and features in step 101, respectively. Therefore, the descriptions of the operations and features in step 101 also apply to step 201, and will not be repeated here.

[0057] Step 202: Send the target image to the intelligent recognition interaction device to obtain the image annotation information of the target image and the recognition requirement information of the corresponding next image output by the intelligent recognition interaction device.

[0058] In this embodiment, the intelligent recognition interactive device is a device that automatically identifies the target type or target attribute type of multiple targets in a target image. The device can provide recognition results for the target type or target attribute type of different targets, and the device can also automatically plan the target type or target attribute type of the target in the next image. The target type or target attribute type automatically planned by the device can be determined by the sorting model in the device (random or output according to the sequence requirements). For example, after the intelligent recognition interactive device recognizes that the action type of the target in the current target image is squatting, it outputs the recognition requirement information of the next image, "Please enter the image of a standing person".

[0059] Step 203: Associate the target image and image annotation information to obtain image samples.

[0060] In this embodiment, the target image in the image sample is associated with the image annotation information. Once the image sample is obtained, the image annotation information can be obtained accordingly.

[0061] In this embodiment, the image annotation information is the information obtained by the intelligent recognition interaction device after recognizing the target image. For example, the image annotation information may include: target type, the probability that the target in the image belongs to the target type; or target attribute type, the probability that the target in the image belongs to the target attribute type.

[0062] In this embodiment, the intelligent recognition interaction device may be a device including an intelligent recognition interaction model. The intelligent recognition interaction model is used to identify targets in a target image and obtain the probability of belonging to a target type or target attribute type.

[0063] Step 204: In response to the fact that the current set of image samples, including image samples, does not meet the requirements of the recognition requirement information, updated visual key information is obtained based on the recognition requirement information.

[0064] It should be understood that the operations and features in step 204 correspond to the operations and features in step 103, respectively. Therefore, the description of the operations and features in step 103 also applies to step 204, and will not be repeated here.

[0065] Optionally, step 204 above includes: searching for image samples corresponding to the recognition requirement information from the current image sample set; determining that the image sample set does not meet the requirements of the recognition requirement information in response to the absence of image samples corresponding to the recognition requirement information in the image sample set; and updating the visual key information based on the recognition requirement information to obtain the updated visual key information.

[0066] Step 205: Based on the updated visual key information and intelligent recognition interaction device, obtain the next image sample, and update the image sample set with the next image sample.

[0067] In this embodiment, step 205 specifically includes: obtaining the next target image based on the updated visual key information; inputting the next target image into the intelligent recognition interaction device to obtain the image annotation information of the next target image; associating the next target image and the image annotation information of the next target image to obtain the next image sample; adding the next image sample to the image sample set to update the image sample set to obtain the current image sample set.

[0068] In this embodiment, the image sample set at the current moment is the current image sample set. As time progresses, the image sample set is automatically updated after the next image sample is added to the image sample set at the next moment, resulting in the image sample set for the next moment. The current image sample set includes current image samples and may also include historical image samples.

[0069] The sample set generation method provided in this disclosure first obtains the current target image based on the current visual key information; second, it sends the target image to an intelligent recognition interaction device to obtain image annotation information of the target image and recognition requirement information for the next image output by the intelligent recognition interaction device; third, it associates the target image and the image annotation information to obtain image samples; thereafter, in response to the current image sample set including image samples not meeting the requirements of the recognition requirement information, it obtains updated visual key information based on the recognition requirement information; finally, it obtains the next image sample based on the updated visual key information and the intelligent recognition interaction device, and updates the image sample set with the next image sample. Therefore, in the process of obtaining the target image through visual key information, the visual key information is updated in real time through the recognition requirement information output by the intelligent recognition interaction device, which allows the image samples corresponding to the target image in the image sample set to gradually increase with the device's needs, automatically expanding the image samples in the target sample set.

[0070] In some optional implementations of this embodiment, obtaining the next image sample based on the updated visual key information and the intelligent recognition interaction device, and updating the image sample set with the next image sample, includes: searching for or generating the next image based on the updated visual key information; inputting the next image into the intelligent recognition interaction device to obtain image annotation information related to the next image output by the intelligent recognition interaction device; associating the next image and the image annotation information related to the next image to obtain the next image sample; adding the next image sample to the image sample set to obtain the updated image sample set; and using the updated image sample set as the current image sample set.

[0071] In this optional implementation, the updated visual key information is the text data of the obtained image. The target image can be searched from an internet search engine using this text data, and the next image can be found. If the next image cannot be found, the next image can be generated using an image generation tool.

[0072] The method for updating the image sample set provided by this optional implementation searches for or generates the next image based on the updated visual key information, and obtains the next image sample from the next image, providing a reliable means for expanding the image sample set and ensuring the reliability of the image sample set expansion.

[0073] Optionally, the above-mentioned method of obtaining the next image sample based on the updated visual key information and the intelligent recognition interaction device, and updating the image sample set with the next image sample, includes: searching for the next image based on the updated visual key information; if no next image is found, generating the next image based on the updated visual key information; inputting the next image into the intelligent recognition interaction device to obtain image annotation information related to the next image output by the intelligent recognition interaction device; associating the next image with the image annotation information related to the next image to obtain the next image sample; adding the next image sample to the image sample set to obtain the updated image sample set; and using the updated image sample set as the current image sample set.

[0074] In some optional implementations of this embodiment, the recognition requirement information includes: action information of the target in the image; the intelligent recognition interaction device includes: a random sorting module and an intelligent interactive recognition model connected to the random sorting module; sending the target image to the intelligent recognition interaction device to obtain the recognition requirement information for the next image output by the intelligent recognition interaction device includes: sending the target image to the intelligent recognition interaction model so that the intelligent recognition interaction model outputs the image annotation information of the target image; controlling the random sorting module to randomly output the action information of the target in the next image after the intelligent recognition interaction model outputs the image annotation information.

[0075] In this optional implementation, the intelligent recognition interaction model is a model for recognizing the actions of targets in an image. This intelligent recognition interaction model is a trained model. For the samples used during the training process, the intelligent recognition interaction model can have a good recognition effect. For actions that do not appear in the samples during the training process, the recognition effect of the intelligent recognition interaction model is uncertain. The image samples obtained by the sample set generation method of this disclosure can provide a reliable prediction basis for the prediction of the intelligent recognition interaction model.

[0076] In this optional implementation, the random sorting module is a module for randomly sorting action information. This module is actually obtained through a script, providing a reliable foundation for guiding the intelligent recognition interactive device to output recognition requirement information in an unordered manner.

[0077] The optional implementation provides a method for obtaining recognition requirement information for the next image. This method randomly sorts actions through a random sorting module in an intelligent recognition interactive device, and after the intelligent recognition interactive model outputs image annotation information, it randomly outputs the action information of the target in the next image, providing a reliable means to expand the image samples in the image sample set.

[0078] Optionally, the intelligent recognition interactive device includes: a sequential sorting module and an intelligent interactive recognition model connected to the sequential sorting module. The sequential sorting module outputs actions sequentially according to a preset action sequence. Sending the target image to the intelligent recognition interactive device and obtaining the recognition requirement information for the next image output by the intelligent recognition interactive device includes: sending the target image to the intelligent recognition interactive model so that the intelligent recognition interactive model outputs image annotation information for the target image; and controlling the random sorting module to output the action information of the target in the next image according to the action sequence after the intelligent recognition interactive model outputs the image annotation information. In this optional implementation, the sequential sorting module is a module that sorts action information in sequence. This module is actually obtained through a script, providing a reliable basis for guiding the intelligent recognition interactive device to output recognition requirement information in an orderly manner.

[0079] Further reference Figure 3 This disclosure provides an embodiment of an automated testing method. Figure 3 A flow 300 is shown as an embodiment of an automated testing method according to the present disclosure, the automated testing method comprising the following steps:

[0080] Step 301: Select image samples related to the action to be tested from the current image sample set to obtain the selected samples.

[0081] In this embodiment, the action to be tested is the action that the intelligent recognition interaction device needs to recognize. The action to be tested can be the action of a target in an image that the intelligent recognition interaction device has not previously encountered.

[0082] In this embodiment, the image sample set is obtained by the sample set generation method provided in the above embodiments. The image samples in the image sample set obtained by the sample set generation method of this disclosure can be obtained by the intelligent recognition interactive device randomly or orderly updating visual key information.

[0083] In this embodiment, in order to better guide the automated testing of intelligent recognition interactive devices, the selected sample from the image sample set can be the image sample in the image sample set that is closest to the action to be tested.

[0084] Step 302: Based on the selected samples, perform automated testing on the intelligent recognition interaction device to obtain the automated test results of the intelligent recognition interaction device.

[0085] In this embodiment, since the intelligent recognition interaction model in the intelligent recognition interaction device recognizes a limited number of images during the training process, and the intelligent recognition interaction device also has multiple modules (such as random sorting module or sequential sorting module), in order to analyze the overall recognition effect of the intelligent recognition interaction device, it is necessary to automatically test the intelligent recognition interaction device and determine the current automated test results of the intelligent recognition interaction device.

[0086] In this embodiment, the automated test results of the intelligent recognition interaction device include one or more of the following: the accuracy of the intelligent recognition interaction device and the reliability of the intelligent recognition interaction device.

[0087] The automated testing method provided in this embodiment selects image samples from the image sample set generated by the sample set generation method, and performs automated testing on the intelligent recognition interactive device based on the selected samples. This can ensure the diversity of intelligent recognition interactive device samples and improve the accuracy of automated testing of intelligent recognition interactive devices.

[0088] In some optional implementations of this embodiment, the above-mentioned selection of image samples related to the action to be tested from the current image sample set to obtain the selected samples includes: obtaining the current image sample set; selecting a sub-sample set of images related to the action to be tested based on the image annotation information of each image sample in the image sample set; and determining the selected samples based on the image annotation information of each image sample in the sub-sample set.

[0089] In this optional implementation, each image sample in the image sample set includes: image annotation information related to various different actions. Specifically, the image annotation information may include: the type of action of the target in the image and the confidence level of that action type.

[0090] The above-mentioned selection of image subsamples related to the action to be tested based on image annotation information includes: determining the action type with the highest confidence in the image annotation information of each image sample, identifying image samples with the same action type as the action to be tested, and combining all image samples with the same action type as the action to be tested together to obtain the image subsample set.

[0091] The above-mentioned determination of the selected samples based on the image annotation information of each image sample in the image subsample set includes: sorting the confidence of each image sample in the image subsample set under the action type from large to small or from small to large, and selecting the image sample with the highest confidence as the selected sample.

[0092] The optional implementation provides a method for obtaining selected samples, which extracts image subsets based on the image annotation information of image samples in the current image sample set; and determines the selected samples based on the image annotation information of the image subsets, thus providing reliable technical support for the extraction of selected samples.

[0093] Optionally, the above-mentioned selection of image samples related to the action to be tested from the current image sample set includes: determining image samples related to the action to be tested based on the action of the target in each image sample in the current image sample set; and selecting any one image sample from the image samples related to the action to be tested as the selection sample.

[0094] In some embodiments of this disclosure, the above-mentioned action to be tested can be obtained through the following steps: obtaining an action sequence from the intelligent recognition interaction device; when performing automated testing on the intelligent recognition interaction device, selecting actions from the action sequence in sequence according to the automated testing order as the action to be tested.

[0095] In this embodiment, the intelligent recognition interaction device includes a sequence sorting module and an intelligent recognition interaction model. The sequence sorting module is used to set an action sequence and, after the intelligent recognition interaction model outputs the recognition result of the image, outputs the recognition requirement information for the next image. The intelligent recognition interaction model is used to represent the correspondence between the target image and the actions of the target in the target image.

[0096] In this optional implementation, the automated test sequence refers to the order in which the intelligent interactive devices are tested. For each pair of intelligent interactive devices tested, an action is selected from the action sequence according to the order indicated by the action sequence.

[0097] The automated testing method provided in this embodiment obtains action sequences from intelligent recognition interactive devices, determines the actions to be tested based on the action sequences, and can perform automated testing of intelligent recognition interactive devices according to the order of actions described in the action sequences, thereby improving the reliability of automated testing of intelligent recognition interactive devices.

[0098] Optionally, the aforementioned intelligent recognition and interaction device may further include: a random sorting module, which is used to randomly sort actions, and the action to be tested may also be an action randomly determined by the random sorting module of the intelligent recognition and interaction device.

[0099] In some optional implementations of this embodiment, the selected sample includes: a target image and image annotation information corresponding to the target image. Based on the selected sample, the intelligent recognition interaction device is automatically tested to obtain the automated test result of the intelligent recognition interaction device, which includes: inputting the target image into the intelligent recognition interaction device to obtain the image annotation information of the action to be tested output by the intelligent recognition interaction device; and determining the accuracy of the intelligent recognition interaction device based on the output image annotation information and the image annotation information of the target image.

[0100] In this optional implementation, the intelligent recognition interaction device includes: an intelligent recognition interaction model, which is used to recognize the actions of targets in an image and output image annotation information. If the image annotation information in a selected sample is exactly the same as the image annotation information output by the intelligent recognition interaction device, or if the similarity between the two is greater than a preset similarity threshold, the accuracy of the intelligent recognition interaction device is determined to be qualified; otherwise, the accuracy of the intelligent recognition interaction device is determined to be unqualified.

[0101] In this embodiment, the image annotation information includes: image format, image type, the action type of the target in the image, and the confidence level that the target belongs to that action type.

[0102] Optionally, determining the accuracy of the intelligent recognition interactive device based on the output image annotation information and the target image image annotation information includes: comparing the image format, image type, and target action type in the output image annotation information with the image format, image type, and target action type in the target image image annotation information; if the image format, image type, and target action type are the same, the confidence scores under each action type are subtracted; if the difference is greater than a preset difference threshold, the accuracy of the intelligent recognition interactive device is determined to be unqualified; if the difference in confidence scores under each action type is less than the difference threshold, the accuracy of the intelligent recognition interactive device is determined to be qualified.

[0103] The optional implementation provides a method for obtaining automated test results of intelligent recognition interactive devices. Based on the image annotation information output by the intelligent recognition interactive device and the image annotation information of selected samples, it determines the accuracy of the intelligent recognition interactive device, providing a reliable means for testing the accuracy of intelligent recognition interactive devices.

[0104] Optionally, the automated test results include: the reliability of the intelligent recognition interactive device; the selected samples include: the target image and the image annotation information corresponding to the target image; based on the selected samples, the intelligent recognition interactive device is automatically tested, and the automated test results of the intelligent recognition interactive device include: inputting the target image into the intelligent recognition interactive device to obtain the recognition requirement information of the next image output by the intelligent recognition interactive device; and determining the reliability of the intelligent recognition interactive device based on the next action adjacent to the action to be tested in the preset action sequence and the recognition requirement information output by the intelligent recognition interactive device.

[0105] In this optional implementation, if the next action and the recognition requirement information output by the intelligent recognition interaction device are the same, the reliability of the intelligent recognition interaction device is determined to be qualified; otherwise, the reliability of the intelligent recognition interaction device is determined to be unqualified.

[0106] Further reference Figure 4 As an implementation of the image set generation method shown in the above figures, this disclosure provides an embodiment of an image set generation apparatus, which is similar to... Figure 1 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0107] like Figure 4 As shown, the image set generation apparatus 400 provided in this embodiment includes: a target image obtaining unit 401, a target information obtaining unit 402, an updating unit 403, and a target set obtaining unit 404. The target image obtaining unit 401 can be configured to obtain the current target image based on the current visual key information. The target information obtaining unit 402 can be configured to send the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device. The updating unit 403 can be configured to obtain updated visual key information based on the recognition requirement information in response to the current target image set including the target image not meeting the requirements of the recognition requirement information. The target set obtaining unit 404 can be configured to obtain the next image based on the updated visual key information and update the target image set using the next image.

[0108] In this embodiment, the specific processing and technical effects of the target image obtaining unit 401, target information obtaining unit 402, updating unit 403, and target set obtaining unit 404 in the image set generation apparatus 400 can be referred to respectively. Figure 1 The relevant descriptions of steps 101, 102, 103, and 104 in the corresponding embodiments will not be repeated here.

[0109] In some optional implementations of this embodiment, the update unit 403 is further configured to: search for target images corresponding to the recognition requirement information from the current target image set; determine that the target image set does not meet the requirements of the recognition requirement information in response to the absence of a target image corresponding to the recognition requirement information in the image sample set; and update the visual key information based on the recognition requirement information to obtain the updated visual key information.

[0110] In some optional implementations of this embodiment, the target set obtaining unit 404 is further configured to: search for or generate the next image based on the updated visual key information; add the next image to the target image set to obtain an updated target image set; and use the updated target image set as the current target image set.

[0111] The image set generation apparatus provided in the embodiments of this disclosure firstly, a target image obtaining unit 401 obtains a current target image based on current visual key information; secondly, a target information obtaining unit 402 sends the target image to an intelligent recognition interaction device to obtain recognition requirement information for the next image output by the intelligent recognition interaction device; thirdly, an updating unit 403, in response to the current target image set including the target image not meeting the requirements of the recognition requirement information, obtains updated visual key information based on the recognition requirement information; finally, a target set obtaining unit 404 obtains the next image based on the updated visual key information and updates the target image set with the next image. Thus, in the process of obtaining the target image through visual key information, the visual key information is updated in real time through the recognition requirement information output by the intelligent recognition interaction device, allowing the target images in the target image set to gradually increase according to the device's needs, automatically expanding the target image set.

[0112] Further reference Figure 5 As an implementation of the sample set generation method shown in the above figures, this disclosure provides an embodiment of a sample set generation apparatus, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0113] like Figure 5As shown, the sample set generation device 500 provided in this embodiment includes: a sample image obtaining unit 501, a recognition unit 502, an association unit 503, a sample information obtaining unit 504, and a sample set obtaining unit 505. The sample image obtaining unit 501 can be configured to obtain the current target image based on the current visual key information. The recognition unit 502 can be configured to send the target image to an intelligent recognition interaction device to obtain image annotation information of the target image output by the intelligent recognition interaction device and recognition requirement information for the corresponding next image. The association unit 503 can be configured to associate the target image and the image annotation information to obtain image samples. The sample information obtaining unit 504 can be configured to obtain updated visual key information based on the recognition requirement information in response to the current image sample set including image samples not meeting the requirements of the recognition requirement information. The sample set obtaining unit 505 can be configured to obtain the next image sample based on the updated visual key information and the intelligent recognition interaction device, and update the image sample set using the next image sample.

[0114] In this embodiment, the specific processing and technical effects of the sample image obtaining unit 501, the recognition unit 502, the association unit 503, the sample information obtaining unit 504, and the sample set obtaining unit 505 in the sample set generation device 500 can be found in the following references. Figure 2 The relevant descriptions of steps 201, 202, 203, 204, and 205 in the corresponding embodiments will not be repeated here.

[0115] In some optional implementations of this embodiment, the sample set obtaining unit 505 is further configured to: search for or generate the next image based on the updated visual key information; input the next image into the intelligent recognition interaction device to obtain image annotation information related to the next image output by the intelligent recognition interaction device; associate the next image and the image annotation information related to the next image to obtain the next image sample; add the next image sample to the image sample set to obtain the updated image sample set; and use the updated image sample set as the current image sample set.

[0116] In some optional implementations of this embodiment, the aforementioned key visual information includes: action information of the target in the image; the intelligent recognition interaction device includes: a random sorting module and an intelligent interactive recognition model connected to the random sorting module; the aforementioned recognition unit 502 is further configured to: send the target image to the intelligent recognition interaction model so that the intelligent recognition interaction model outputs image annotation information of the target image; and control the random sorting module to randomly output the action information of the target in the next image after the intelligent recognition interaction model outputs the image annotation information.

[0117] Further reference Figure 6 As an implementation of the automated testing methods shown in the above figures, this disclosure provides an embodiment of an automated testing apparatus, which is similar to... Figure 3 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0118] like Figure 6 As shown, the automated testing device 600 provided in this embodiment includes a selection unit 601 and a testing unit 602. The selection unit 601 can be configured to select image samples related to the action to be tested from the current image sample set to obtain selected samples; the image sample set is obtained based on the sample set generation device in the above embodiment. The testing unit 602 can be configured to perform automated testing on the intelligent recognition interactive device based on the selected samples to obtain the automated test results of the intelligent recognition interactive device.

[0119] In this embodiment, in the automated testing device 600: the specific processing of the selected unit 601 and the testing unit 602, and the resulting technical effects, can be referred to respectively. Figure 3 The relevant descriptions of steps 301 and 302 in the corresponding embodiments will not be repeated here.

[0120] In some optional implementations of this embodiment, the selection unit 601 is configured to: obtain the current image sample set, wherein each image sample in the image sample set includes: image annotation information related to multiple different actions; select an image sub-sample set related to the action to be tested based on the image annotation information of each image sample in the image sample set; and determine the selected sample based on the image annotation information of each image sample in the image sub-sample set.

[0121] In some optional implementations of this embodiment, the action to be tested is obtained by an action determination module (not shown in the figure). The action determination module can be configured to obtain an action sequence from the intelligent recognition interaction device. When performing automated testing on the intelligent recognition interaction device, actions are selected sequentially from the action sequence according to the automated testing order as the action to be tested.

[0122] In some optional implementations of this embodiment, the selected sample includes: a target image and image annotation information corresponding to the target image. The test unit 601 is further configured to: input the target image into the intelligent recognition interaction device to obtain the image annotation information of the action to be tested output by the intelligent recognition interaction device; and determine the accuracy of the intelligent recognition interaction device based on the output image annotation information and the image annotation information of the target image.

[0123] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0124] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0125] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0126] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded into random access memory (RAM) 703 from storage unit 708. RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0127] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0128] The computing unit 701 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as image set generation methods, sample set generation methods, or automated testing methods. For example, in some embodiments, the image set generation method, sample set generation method, or automated testing method can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the image set generation method, sample set generation method, or automated testing method described above can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured by any other suitable means (e.g., by means of firmware) to perform an image set generation method, a sample set generation method, or an automated testing method.

[0129] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0130] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to the processor or controller of a general-purpose computer, special-purpose computer, or other programmable image set generation apparatus, sample set generation apparatus, or automated testing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0131] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0133] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0134] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0135] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0136] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for generating an image set, the method comprising: Based on the current key visual information, the current target image is obtained, wherein the key visual information includes: image format, image type, image label and / or attribute information of the target in the image; The target image is sent to the intelligent recognition interaction device to obtain the recognition requirement information for the next image output by the intelligent recognition interaction device, wherein the recognition requirement information includes the target type or the type of target attribute; In response to the current set of target images including the target image not meeting the requirements of the recognition requirement information, updated visual key information is obtained based on the recognition requirement information; Based on the updated visual key information, the next image is searched or generated, and the target image set is updated using the next image.

2. The method according to claim 1, wherein, In response to the current set of target images including the target image not meeting the requirements of the recognition requirement information, the updated visual key information obtained based on the recognition requirement information includes: Search the current set of target images for the target image that corresponds to the recognition requirement information; In response to the fact that there is no target image in the image sample set that corresponds to the recognition requirement information, it is determined that the target image set does not meet the requirements of the recognition requirement information; Based on the identification requirement information, the visual key information is updated to obtain the updated visual key information.

3. The method according to claim 1, wherein, The step of obtaining the next image based on the updated visual key information and updating the target image set using the next image includes: Based on the updated visual key information, search for or generate the next image; Add the next image to the target image set to obtain an updated target image set; The updated target image set is used as the current target image set.

4. A method for generating a sample set, the method comprising: Based on the current key visual information, the current target image is obtained, wherein the key visual information includes: image format, image type, image label and / or attribute information of the target in the image; The target image is sent to the intelligent recognition interaction device to obtain the image annotation information of the target image and the recognition requirement information of the corresponding next image output by the intelligent recognition interaction device, wherein the recognition requirement information includes the target type or the type of target attribute; By associating the target image with the image annotation information, an image sample is obtained; In response to the current image sample set including the image sample not meeting the requirements of the recognition requirement information, updated visual key information is obtained based on the recognition requirement information; Based on the updated visual key information and the intelligent recognition interaction device, the next image sample is searched or generated, and the image sample set is updated using the next image sample.

5. The method according to claim 4, wherein, The step of obtaining the next image sample based on the updated visual key information and the intelligent recognition interaction device, and updating the image sample set with the next image sample, includes: Based on the updated visual key information, search for or generate the next image; The next image is input into the intelligent recognition interaction device to obtain image annotation information related to the next image output by the intelligent recognition interaction device; By associating the next image with the image annotation information associated with the next image, a next image sample is obtained; Add the next image sample to the image sample set to obtain an updated image sample set; Use the updated image sample set as the current image sample set.

6. The method according to claim 4, wherein, The recognition requirement information includes: action information of the target in the image; the intelligent recognition interaction device includes: a random sorting module and an intelligent interactive recognition model connected to the random sorting module; the step of sending the target image to the intelligent recognition interaction device and obtaining the recognition requirement information of the next image output by the intelligent recognition interaction device includes: The target image is sent to the intelligent recognition interaction model so that the intelligent recognition interaction model outputs the image annotation information of the target image; After the intelligent recognition interaction model outputs the image annotation information, the random sorting module is controlled to randomly output the action information of the target in the next image.

7. An automated testing method, the method comprising: Select image samples related to the action to be tested from the current image sample set to obtain the selected samples; The image sample set is obtained based on the sample set generation method according to any one of claims 4-6; Based on the selected samples, the intelligent recognition interaction device is subjected to automated testing to obtain the automated test results of the intelligent recognition interaction device.

8. The method according to claim 7, wherein, The step of selecting image samples related to the action to be tested from the current image sample set includes: Obtain the current image sample set, wherein each image sample in the image sample set includes: image annotation information related to multiple different actions; Based on the image annotation information of each image sample in the image sample set, a subset of images related to the action to be tested is selected; Based on the image annotation information of each image sample in the image subsample set, the selected samples are determined.

9. The method according to claim 7, wherein, The selected samples include: a target image and image annotation information corresponding to the target image. Based on the selected samples, the intelligent recognition interaction device is subjected to automated testing, and the automated test results of the intelligent recognition interaction device include: The target image is input into the intelligent recognition and interaction device to obtain the image annotation information of the action to be tested output by the intelligent recognition and interaction device; The accuracy of the intelligent recognition interactive device is determined based on the output image annotation information and the image annotation information of the target image.

10. An image set generation apparatus, the apparatus comprising: The target image obtaining unit is configured to obtain the current target image based on the current visual key information, wherein the visual key information includes: image format, image type, image label and / or attribute information of the target in the image; The target information obtaining unit is configured to send the target image to the intelligent recognition interaction device to obtain the recognition requirement information of the next image output by the intelligent recognition interaction device, wherein the recognition requirement information includes the target type or the type of target attribute; The update unit is configured to, in response to the current set of target images including the target image not meeting the requirements of the recognition requirement information, obtain updated visual key information based on the recognition requirement information; The target set obtaining unit is configured to search for or generate the next image based on the updated visual key information, and update the target image set with the next image.

11. The apparatus according to claim 10, wherein, The updating unit is further configured to: search for a target image corresponding to the recognition requirement information from the current target image set; and determine that the target image set does not meet the requirements of the recognition requirement information in response to the absence of a target image corresponding to the recognition requirement information in the image sample set. Based on the identification requirement information, the visual key information is updated to obtain the updated visual key information.

12. The apparatus according to claim 10, wherein, The target set obtaining unit is further configured to: search for or generate the next image based on the updated visual key information; add the next image to the target image set to obtain an updated target image set; and use the updated target image set as the current target image set.

13. A sample set generation apparatus, the apparatus comprising: The sample image obtaining unit is configured to obtain the current target image based on the current visual key information, wherein the visual key information includes: image format, image type, image label and / or attribute information of the target in the image; The recognition unit is configured to send the target image to the intelligent recognition interaction device to obtain the image annotation information of the target image and the recognition requirement information of the corresponding next image output by the intelligent recognition interaction device, wherein the recognition requirement information includes the target type or the type of the target attribute; The association unit is configured to associate the target image and the image annotation information to obtain an image sample; The sample information obtaining unit is configured to obtain updated visual key information based on the recognition requirement information in response to the current image sample set including the image sample not meeting the requirements of the recognition requirement information. The sample set obtaining unit is configured to search for or generate the next image sample based on the updated visual key information and the intelligent recognition interaction device, and update the image sample set with the next image sample.

14. The apparatus according to claim 13, wherein, The sample set obtaining unit is further configured to: search for or generate the next image based on the updated visual key information; input the next image into the intelligent recognition interaction device to obtain image annotation information related to the next image output by the intelligent recognition interaction device; Associate the next image with the image annotation information associated with the next image to obtain the next image sample; add the next image sample to the image sample set to obtain the updated image sample set; use the updated image sample set as the current image sample set.

15. The apparatus according to claim 13, wherein, The key visual information includes: action information of the target in the image; the intelligent recognition interaction device includes: a random sorting module and an intelligent interactive recognition model connected to the random sorting module; the recognition unit is further configured to: send the target image to the intelligent recognition interaction model so that the intelligent recognition interaction model outputs image annotation information of the target image; and control the random sorting module to randomly output the action information of the target in the next image after the intelligent recognition interaction model outputs the image annotation information.

16. An automated testing apparatus, the apparatus comprising: The selection unit is configured to select image samples related to the action to be measured from the current image sample set to obtain the selected samples; The image sample set is obtained based on the sample set generation apparatus according to any one of claims 13-15; The testing unit is configured to perform automated testing on the intelligent recognition interaction device based on the selected samples, and obtain the automated test results of the intelligent recognition interaction device.

17. The apparatus according to claim 16, wherein, The selection unit is configured to: acquire the current image sample set, wherein each image sample in the image sample set includes image annotation information related to multiple different actions; select an image sub-sample set related to the action to be tested based on the image annotation information of each image sample in the image sample set; and determine the selected sample based on the image annotation information of each image sample in the image sub-sample set.

18. The apparatus according to claim 16, wherein, The selected samples include: a target image and image annotation information corresponding to the target image. The test unit is further configured to: input the target image into the intelligent recognition interaction device to obtain the image annotation information of the action to be tested output by the intelligent recognition interaction device; and determine the accuracy of the intelligent recognition interaction device based on the output image annotation information and the image annotation information of the target image.

19. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-9.

20. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-9.

21. A computer program product comprising a computer program that, when executed by a processor, implements the method of any one of claims 1-9.

Images