Object identification detection method, device, equipment and readable storage medium

Abstract

The present application discloses an object recognition detection method, device, equipment and readable storage medium, the object recognition detection method comprising: obtaining time frame images taken by each preset camera in each time frame, and based on the transformation relationship between each preset camera, determining the object correspondence between each time frame image corresponding to each time frame, and then determining the target detection object corresponding to the object correspondence, and calculating the quality index value of each time frame image for the target detection object, and then based on each quality index value, determining key frames in each time frame image corresponding to each preset camera, and then based on each key frame, identifying and detecting the target detection object to obtain a target recognition detection result. The present application solves the technical problem of low accuracy of object recognition detection.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence in financial technology (Fintech), and in particular to an object identification and detection method, device, equipment and readable storage medium. Background Art

[0002] With the continuous development of financial technology, especially Internet technology finance, more and more technologies (such as distributed, blockchain, artificial intelligence, etc.) are applied in the financial field, but the financial industry also puts forward higher requirements on technology, such as higher requirements on the distribution of corresponding to-do items in the financial industry.

[0003] At present, in many fields, it is necessary to identify and detect the target detection object in the image. For example, in the intelligent warehouse management system, it is necessary to take the image of the goods and identify the quantity and category of the goods in the image. To identify the target detection object in the image, the image shooting angle needs to be appropriate and the target detection object in the image is clearly visible, so that a good recognition result can be obtained. However, in actual application scenarios, due to many factors, it is difficult to shoot the target detection object in a static state, and the target detection object can only be captured when it passes through the camera. However, due to various uncertain factors such as the movement speed of the target detection object, changes in ambient light, and occlusion of unknown objects, most of the captured images have problems such as unclearness and inappropriate angles, and are not suitable for identifying the target detection object. Therefore, it is currently difficult to select the most suitable image (key frame) based on the captured image for target detection object recognition, resulting in a low accuracy rate for the recognition and detection of the target detection object. Summary of the invention

[0004] The main purpose of the present application is to provide an object identification detection method, device, equipment and readable storage medium, aiming to solve the technical problem of low accuracy of object identification detection in the prior art.

[0005] To achieve the above objectives, the present application provides an object recognition detection method, the object recognition detection method comprising:

[0006] Acquire time frame images taken by each preset camera in each time frame, and determine the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera;

[0007] Determine the target detection object corresponding to the object correspondence relationship, and calculate the quality index value of each of the time frame images for the target detection object;

[0008] Based on the quality index values, key frames are determined in the time frame images corresponding to the preset cameras respectively;

[0009] Based on each of the key frames, the target detection object is identified and detected to obtain a target recognition detection result.

[0010] Optionally, the transformation relationship includes a coordinate transformation relationship,

[0011] The step of determining the object correspondence between the time frame images corresponding to the time frames based on the transformation relationship between the preset cameras comprises:

[0012] Selecting a reference camera from each of the preset cameras, and determining a reference coordinate system corresponding to the reference camera;

[0013] Based on the coordinate conversion relationship, the to-be-determined detection object in each of the time frame images is converted into the reference coordinate system, and similarity index values ​​between each of the to-be-determined detection objects are calculated in the reference coordinate system;

[0014] Based on the similarity index values, the object correspondence is determined.

[0015] Optionally, the step of calculating the similarity index values ​​between each pair of the to-be-determined detection objects in the reference coordinate system comprises:

[0016] Determine the intersection and union of each of the to-be-determined detection objects in the reference coordinate system;

[0017] Based on the intersection parts and the union parts, the similarity index values ​​are calculated.

[0018] Optionally, the step of performing recognition detection on the target detection object based on each of the key frames to obtain a target recognition detection result includes:

[0019] Input each of the key frames into a preset object recognition and detection model, respectively, perform recognition and detection on the target detection object in each of the key frames, and obtain recognition and detection results corresponding to each of the key frames;

[0020] Based on the key frame confidence corresponding to each of the key frames, a target recognition detection result is determined in each of the recognition detection results.

[0021] Optionally, the step of calculating the quality index value of each of the time frame images for the target detection object includes:

[0022] Respectively identifying each of the time frame images to obtain key information of the contour of the target detection object in each of the time frame images;

[0023] According to the area defined by each of the contour key information, the quality index value of each of the time frame images is calculated respectively.

[0024] Optionally, the contour key information includes key point information, and the quality index value includes a contour surface index value and a distance index value.

[0025] The step of respectively calculating the quality index value of each time frame image according to the area defined by each contour key information comprises:

[0026] According to the key point information, respectively determine the contour surface area formed by each key point in each time frame image and the actual position of the target detection object in each time frame image;

[0027] Respectively calculating the profile surface index value of each profile surface area under a preset profile surface quality index;

[0028] The distance between each of the actual positions and the preset optimal position is calculated respectively to obtain the distance index value corresponding to each of the time frame images.

[0029] Optionally, the quality index value includes a contour surface index value and a distance index value, and each of the time frame images includes a current time frame image.

[0030] The step of determining a key frame in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values ​​comprises:

[0031] Determine the current key frame of each preset camera, and respectively obtain the key frame contour surface index value and the key frame distance index value corresponding to each current key frame;

[0032] Respectively detecting whether the distance ratio between the distance index value of each current time frame image and the corresponding key frame distance index value is less than a preset ratio, and detecting whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value;

[0033] If the distance ratio is less than the preset ratio and the contour surface index value is greater than the key frame contour surface index value, the current time frame image is updated to the current key frame until all the time frame images are detected and each key frame is obtained.

[0034] The present application also provides an object recognition and detection device, which is a virtual device and includes:

[0035] An acquisition module, used to acquire time frame images taken by each preset camera in each time frame, and determine the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera;

[0036] A first determination module is used to determine the target detection object corresponding to the object correspondence relationship, and calculate the quality index value of each of the time frame images for the target detection object;

[0037] A second determination module is used to determine key frames in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values;

[0038] The recognition and detection module is used to recognize and detect the target detection object based on each of the key frames to obtain a target recognition and detection result.

[0039] Optionally, the acquisition module includes:

[0040] A selection unit, configured to select a reference camera from among the preset cameras and determine a reference coordinate system corresponding to the reference camera;

[0041] A conversion unit, configured to convert the to-be-determined detection object in each of the time frame images into the reference coordinate system based on the coordinate conversion relationship, and calculate similarity index values ​​between each of the to-be-determined detection objects in the reference coordinate system;

[0042] The first determining unit is used to determine the object correspondence based on each of the similarity index values.

[0043] Optionally, the conversion unit includes:

[0044] A first determination subunit is used to determine the intersection part and the union part between each of the to-be-determined detection objects in the reference coordinate system;

[0045] The first calculation subunit is used to calculate each of the similarity index values ​​based on each of the intersection parts and each of the union parts.

[0046] Optionally, the identification and detection module includes:

[0047] An identification and detection unit, used to input each of the key frames into a preset object identification and detection model, respectively, to identify and detect the target detection object in each of the key frames, and obtain an identification and detection result corresponding to each of the key frames;

[0048] The second determination unit is used to determine the target recognition detection result in each of the recognition detection results based on the key frame confidence corresponding to each of the key frames.

[0049] Optionally, the first determining module includes:

[0050] An identification unit, used to identify each of the time frame images respectively, and obtain key information of the contour of the target detection object in each of the time frame images;

[0051] The calculation unit is used to calculate the quality index value of each time frame image according to the area defined by each contour key information.

[0052] Optionally, the computing unit includes:

[0053] A second determination subunit is used to determine the contour surface area formed by each key point in each time frame image and the actual position of the target detection object in each time frame image according to each key point information;

[0054] A second calculation subunit, used to respectively calculate the contour surface index value of each contour surface area under a preset contour surface quality index;

[0055] The third calculation subunit is used to respectively calculate the distance between each of the actual positions and the preset optimal position, and obtain the distance index value corresponding to each of the time frame images.

[0056] Optionally, the second determining module includes:

[0057] An acquisition unit, used to determine the current key frame of each preset camera, and respectively acquire the key frame contour surface index value and the key frame distance index value corresponding to each current key frame;

[0058] A detection unit, used to respectively detect whether the distance ratio between the distance index value of each current time frame image and the corresponding key frame distance index value is less than a preset ratio, and to detect whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value;

[0059] An updating unit is used to update the current time frame image to the current key frame if the distance ratio is less than the preset ratio and the contour surface index value is greater than the key frame contour surface index value, until all the time frame images are detected and all the key frames are obtained.

[0060] The present application also provides an object identification and detection device, which is a physical device, and includes: a memory, a processor, and a program of the object identification and detection method stored in the memory and executable on the processor. When the program of the object identification and detection method is executed by the processor, the steps of the object identification and detection method as described above can be implemented.

[0061] The present application also provides a readable storage medium, on which is stored a program for implementing the object identification and detection method. When the program of the object identification and detection method is executed by a processor, the steps of the object identification and detection method as described above are implemented.

[0062] The present application obtains the time frame images taken by each preset camera in each time frame, and based on the transformation relationship between each preset camera, determines the object correspondence relationship between each time frame image corresponding to each time frame, and then determines the target detection object corresponding to the object correspondence relationship, and calculates the quality index value of each time frame image for the target detection object, and then based on each quality index value, determines the key frames in each time frame image corresponding to each preset camera, and then based on each key frame, identifies and detects the target detection object to obtain a target recognition detection result. That is, the present application provides an object recognition and detection method based on multiple cameras, that is, after obtaining the time frame images taken by each preset camera in each time frame, the object correspondence between each time frame image corresponding to each time frame is determined based on the conversion relationship between each preset camera, wherein, since each preset camera can be arranged at different positions and angles to shoot the target detection object, the probability of poor image quality caused by factors such as object obstruction and poor ambient light when capturing the target detection object is reduced, the overall quality of the captured image is improved, and the image quality used for object recognition and detection is improved, thereby determining The target detection object corresponding to the object correspondence relationship, and based on the calculated quality index values, the key frames are determined in the time frame images corresponding to the preset cameras respectively, that is, the purpose of selecting high-quality key frames in the time frame images corresponding to the preset cameras respectively with image quality as an indicator is achieved, and the image quality used for object recognition detection is further improved, and then the recognition detection of the target detection object is performed based on the key frames, which can achieve the purpose of object recognition detection based on high-quality images, thereby improving the accuracy of object recognition detection, and therefore, solving the technical problem of low accuracy of object recognition detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0064] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0065] Figure 1 This is a flow chart of the first embodiment of the object identification and detection method of the present application;

[0066] Figure 2This is a flow chart of the second embodiment of the object identification and detection method of the present application;

[0067] Figure 3 This is a flow chart of the third embodiment of the object identification and detection method of the present application;

[0068] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present application.

[0069] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0070] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0071] The present application provides an object recognition detection method. In the first embodiment of the object recognition detection method of the present application, refer to Figure 1 , the object identification and detection method comprises:

[0072] Step S10, obtaining time frame images taken by each preset camera in each time frame, and determining the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera;

[0073] In this embodiment, it should be noted that the time frame image is an image of the target detection object taken by the preset camera in the corresponding time frame, wherein the target detection object is an item that needs to be identified and detected, and the time frame image includes at least one target detection object, and each of the preset cameras is arranged at different shooting positions and shooting angles.

[0074] In addition, it should be noted that the transformation relationship is the posture relationship between the preset cameras. For example, assuming that the camera coordinate system of the preset camera A is C and the camera coordinate system of the preset camera B is W, then Xc=R*Xw+t, Yc=R*Yw+t, Zc=R*Zw+t, wherein Xc is the X coordinate of the camera coordinate system C, Yc is the Y coordinate of the camera coordinate system C, Zc is the Z coordinate of the camera coordinate system C, Xw is the X coordinate of the camera coordinate system W, Yw is the Y coordinate of the camera coordinate system W, Zw is the Z coordinate of the camera coordinate system W, R represents the rotation between the camera coordinate system C and the camera coordinate system W, and t represents the translation between the camera coordinate system C and the camera coordinate system W.

[0075] The time frame images taken by each preset camera at each time frame are obtained, and based on the transformation relationship between each preset camera, the object correspondence relationship between each time frame image corresponding to each time frame is determined. Specifically, the time frame images taken by each preset camera at each time frame are obtained, and a reference camera is selected from each preset camera, and the camera coordinate system corresponding to the reference camera is used as the reference coordinate system, wherein the reference coordinate system includes a three-dimensional space coordinate system and a plane coordinate system. For example, the contact surface between the object to be detected and the ground is taken to establish the plane coordinate system, etc., so as to determine the time frame images taken by each preset camera corresponding to each time frame, and for each time frame, the following steps are performed:

[0076] The objects to be confirmed in each of the time frame images corresponding to the time frame are respectively converted into the reference coordinate system, and then the similarity index values ​​between the objects to be confirmed are calculated, and based on the similarity index values, the one-to-one correspondence between the objects to be confirmed is determined, that is, the object correspondence corresponding to each set of images to be identified is obtained, wherein the objects to be confirmed are objects suspected to be target objects, each of the time frame images includes at least one object to be confirmed, and the similarity index values ​​include IoU values ​​(Intersection over Union, the intersection ratio of the detection result area and the true value area), etc.

[0077] Wherein, the transformation relationship includes a coordinate transformation relationship,

[0078] The step of determining the object correspondence between the time frame images corresponding to the time frames based on the transformation relationship between the preset cameras comprises:

[0079] Step S21, selecting a reference camera from the preset cameras, and determining a reference coordinate system corresponding to the reference camera;

[0080] In this embodiment, it should be noted that the reference coordinate system is the camera coordinate system of the reference camera, wherein the camera coordinate system includes a three-dimensional space coordinate system and a plane coordinate system, etc.

[0081] Step S22, based on the coordinate conversion relationship, converting the to-be-determined detection object in each of the time frame images into the reference coordinate system, and calculating the similarity index values ​​between each of the to-be-determined detection objects in the reference coordinate system;

[0082] In this embodiment, it should be noted that the coordinate transformation relationship is a transformation relationship between the plane coordinate systems of the shooting ground surfaces corresponding to the preset cameras.

[0083] Based on the coordinate transformation relationship, the objects to be determined in each of the time frame images are transformed into the reference coordinate system, and the similarity index values ​​between each of the objects to be determined are calculated in the reference coordinate system. Specifically, based on the coordinate transformation relationship, the bottom surface of the objects to be determined in each of the time frame images corresponding to each of the time frames is transformed into the reference coordinate system to obtain the transformation position of each of the objects to be determined in the reference coordinate system. Furthermore, the transformation positions corresponding to each of the time frames are compared respectively to calculate the similarity index values ​​between each of the objects to be determined corresponding to each of the time frames.

[0084] Step S23: determining the object correspondence based on the similarity index values.

[0085] In this embodiment, it should be noted that the time frame image may include multiple objects to be determined, and the object correspondence is a one-to-one correspondence between the objects to be determined in each of the time frame images. If the time frame image taken by the preset camera A in the current time frame includes object A and object B, and the time frame image taken by the preset camera B in the current time frame includes object A1 and object B1, and the similarity index value between object A and object A1 is 0.99, and the similarity index value between object A and object B1 is 0.2, then it is determined that object A corresponds to object A1 one-to-one, and object B corresponds to object B1 one-to-one, that is, the object correspondence is obtained, that is, it is determined that object A and object A1 belong to the same target detection object, and object B and object B1 belong to the same target detection object.

[0086] Based on each of the similarity index values, the object correspondence is determined. Specifically, it is judged whether each of the similarity index values ​​in each of the time frames reaches a preset similarity condition. If the similarity index value reaches the preset similarity condition, it is determined that the two to-be-determined detection objects corresponding to the similarity index value are the same target detection object, and then the to-be-determined detection object corresponding to each of the target detection objects in each of the time frame images is determined, that is, a one-to-one correspondence between each to-be-determined detection object in each of the time frame images corresponding to each time frame is obtained, that is, the correspondence between each of the objects is obtained, wherein the preset similarity condition includes that the IoU values ​​of the two to-be-determined detection objects for a preset number of consecutive time frames are both greater than or equal to a preset IoU value threshold.

[0087] The step of calculating the similarity index values ​​between each pair of the to-be-determined detection objects in the reference coordinate system comprises:

[0088] Step S231, determining the intersection and union of each pair of the to-be-determined detection objects in the reference coordinate system;

[0089] In this embodiment, it should be noted that the intersection part is the intersection volume between the objects to be determined or the intersection area with the ground contact surface, and the union part is the union volume between the objects to be determined or the union area with the ground contact surface.

[0090] Determine the intersection parts and union parts between each of the objects to be detected in the reference coordinate system. Specifically, calculate the intersection areas and union areas between each of the bottom surfaces of the objects to be detected in the reference coordinate system that are in contact with the ground, and obtain each intersection area and each corresponding union area, wherein one intersection area corresponds to one union area.

[0091] Step S232: Calculate each of the similarity index values ​​based on each of the intersection parts and each of the union parts.

[0092] In this embodiment, each similarity index value is calculated based on each intersection part and each union part. Specifically, the ratio between each intersection area and the corresponding union area is calculated, and each ratio is used as the corresponding similarity index value.

[0093] Step S20, determining the target detection object corresponding to the object correspondence relationship, and calculating the quality index value of each of the time frame images for the target detection object;

[0094] In this embodiment, the target detection object corresponding to the object correspondence relationship is determined, and the quality index value of each time frame image for the target detection object is calculated. Specifically, the target detection object corresponding to the object correspondence relationship is determined, and the following steps are performed for each time frame image: based on a preset target detection model, the time frame image is identified to obtain contour key information of the target detection object in the time frame image, and then the quality index value of each time frame image is obtained according to the contour surface parameters of the contour surface area defined by each contour key information, wherein the contour surface parameters include contour surface area, contour surface clarity and contour surface visibility, etc.

[0095] Step S30, based on each of the quality indicator values, determining key frames in each of the time frame images corresponding to each of the preset cameras;

[0096] In this embodiment, based on each of the quality indicator values, key frames are determined in each of the time frame images corresponding to each of the preset cameras. Specifically, based on the preset quality conditions, target quality indicator values ​​are selected from each of the quality indicator values ​​corresponding to each of the preset cameras, and the time frame images corresponding to the target instruction indicator values ​​are used as key frames corresponding to each of the preset cameras, wherein the key frames are time frame images used for object identification and detection.

[0097] Step S40: Based on each of the key frames, the target detection object is identified and detected to obtain a target identification and detection result.

[0098] In this embodiment, based on each of the key frames, the target detection object is identified and detected to obtain a target recognition detection result. Specifically, each of the key frames is respectively input into a preset recognition detection model, and the image area of ​​the target detection object in the key frame is respectively identified and detected to obtain the recognition detection result corresponding to each of the key frames. Then, based on each of the recognition detection results, the target recognition detection result corresponding to the target detection object is determined, wherein the recognition detection result includes recognition detection results of the quantity and category of the target detection object, etc. For example, assuming that there are 10 recognition detection results, among which 9 of the recognition detection results indicate that the target detection object is an apple, and 1 of the recognition detection results indicates that the target detection object is a pear, then the target recognition detection result is that there is a 90% probability that the target detection object is an apple, and a 10% probability that it is a pear.

[0099] Wherein, the step of identifying and detecting the target detection object corresponding to each of the to-be-identified image sets based on each of the key frames includes:

[0100] Step S41, inputting each of the key frames into a preset object recognition and detection model, respectively, performing recognition and detection on the target detection object in each of the key frames, and obtaining recognition and detection results corresponding to each of the key frames;

[0101] In this embodiment, each of the key frames is respectively input into a preset object recognition and detection model, and the target detection object in each of the key frames is respectively recognized and detected to obtain recognition and detection results corresponding to each of the key frames. Specifically, each of the key frames is respectively input into a preset recognition and detection model, and a preset number of convolution and pooling alternating processes are performed on the pixel matrix corresponding to each of the key frames to obtain the convolution and alternating processing results corresponding to each of the key frames. The convolution and alternating processing results are fully connected to obtain the recognition and detection vector corresponding to each of the key frames, wherein the recognition and detection vector includes all feature information in the key frames, and then the recognition and detection results are respectively extracted from each of the recognition and detection vectors.

[0102] Step S42, determining the target recognition detection result from each of the recognition detection results based on the key frame confidence corresponding to each of the key frames.

[0103] In this embodiment, it should be noted that the key frame confidence is the reliability of the key frame.

[0104] Based on the key frame confidence corresponding to each of the key frames, the target recognition detection result is determined in each of the recognition detection results. Specifically, based on the contour surface parameters corresponding to the key frames, the key frame confidences corresponding to each of the key frames are calculated respectively by a preset key frame confidence calculation formula, and the recognition detection result with the highest key frame confidence is selected as the target recognition detection result, wherein the preset key frame confidence calculation formula is as follows:

[0105] G = sum(S i *C i *V i )

[0106] Among them, G is the key frame confidence, Si is the area of ​​the i-th surface of the target detection object, Ci is the visibility of the i-th surface of the target detection object, and Vi is the clarity of the i-th surface of the target detection object.

[0107] This embodiment obtains the time frame images taken by each preset camera in each time frame, and based on the transformation relationship between each preset camera, determines the object correspondence relationship between each time frame image corresponding to each time frame, and then determines the target detection object corresponding to the object correspondence relationship, and calculates the quality index value of each time frame image for the target detection object, and then based on each quality index value, determines the key frame in each time frame image corresponding to each preset camera, and then based on each key frame, identifies and detects the target detection object to obtain a target recognition detection result. That is, this embodiment provides an object recognition and detection method based on multiple cameras, that is, after obtaining the time frame images taken by each preset camera in each time frame, the object correspondence between each time frame image corresponding to each time frame is determined based on the conversion relationship between each preset camera, wherein, since each preset camera can be arranged at different positions and angles to shoot the target detection object, the probability of poor image quality caused by factors such as object obstruction and poor ambient light when capturing the target detection object is reduced, the overall quality of the captured image is improved, and the image quality used for object recognition and detection is improved, thereby ensuring The target detection object corresponding to the object correspondence relationship is determined, and based on the calculated quality index values, the key frames are determined in the time frame images corresponding to the preset cameras respectively. That is, the purpose of selecting high-quality key frames in the time frame images corresponding to the preset cameras respectively with image quality as an indicator is achieved, and the image quality used for object recognition detection is further improved. Then, based on the key frames, recognition detection of the target detection object is performed, which can achieve the purpose of object recognition detection based on high-quality images, thereby improving the accuracy of object recognition detection. Therefore, the technical problem of low accuracy of object recognition detection is solved.

[0108] Further, refer to Figure 2 Based on the first embodiment of the present application, in another embodiment of the present application, the step of calculating the quality index value of each of the time frame images for the target detection object includes:

[0109] Step S21, identifying each of the time frame images respectively to obtain key information of the contour of the target detection object in each of the time frame images;

[0110] In this embodiment, after obtaining the time frame image corresponding to each of the preset cameras, it is determined whether to use the time frame image as a key frame, wherein the key frame is a determined image suitable for identifying and detecting the target detection object. Among them, the target detection object is different in combination with the specific application scenario. For example, in intelligent warehousing, it is necessary to identify the quantity and category of incoming and outgoing goods. In this application scenario, the target detection object is goods, and the time frame image can be an image of goods taken by a camera device set up in an intelligent storage place. Then in other application scenarios, the target detection object can also be a vehicle, a face or vegetation, etc.

[0111] After obtaining each time frame image, each of the time frame images can be identified respectively to obtain the key information of the outline of the target detection object in each of the time frame images. Wherein, the key information of the outline can include the information of the key features related to the outline such as the outline points and contour lines of the target detection object, such as the position of the key features in the image, whether it is visible, etc. A target detection model can be set in advance, and the key information of the outline of the target detection object in the time frame image can be identified by this target detection model. This target detection model can be a target detection model commonly used in the existing, such as Yolov3 network (a target detection network), and a key information detection branch is added to the target detection model, and a target detection model capable of identifying the key information of the outline of the target detection object in the time frame image is obtained by pre-training the target detection model. For example, images with goods can be collected in large quantities, and each image is pre-marked with the key information of the outline of the goods, and these images are used to train the target detection model, and the accuracy of the model recognition key information of the outline is supervised by the marking information, so as to obtain the target detection model capable of identifying the key information of the outline of the goods.

[0112] It should be noted that the time frame image may include multiple target detection objects, and the contour key features that are farther away can be divided by the position of the contour key features in the contour key information, so as to distinguish the contour key information of multiple target detection objects. For one of the target detection objects, according to the subsequent processing flow, the contour key information of the target detection object is used to determine whether the time frame image can be used as a key frame for identifying and detecting the target detection object. In other words, for a time frame image, the time frame image may include multiple target detection objects, and the time frame image may be suitable as a key frame for a part of the target detection objects, but not suitable as a key frame for another part of the target detection objects.

[0113] Step S22, respectively calculating the quality index value of each of the time frame images according to the area defined by each of the contour key information;

[0114] In this embodiment, after obtaining the contour key information of the target detection object in the time frame image, the quality index value of the time frame image can be calculated according to the contour key information. Among them, some quality indicators for evaluating whether the time frame image is suitable as a key frame can be pre-set, and there can be multiple quality indicators, for example, they can include indicators such as clarity and visibility. The quality index value is calculated based on the contour key information, and specifically, the quality index value can be calculated based on the area defined by the contour key information. For example, when the quality index includes clarity, the clarity of the area defined by the contour key information in the time frame image can be calculated. The calculation method of clarity can adopt the existing image clarity calculation method, which will not be described in detail here. There are many ways to determine the area defined by the contour key information. For example, according to the position of each contour key feature in the contour key information, each contour key feature at the edge of each contour key feature can be connected to obtain an area, and the area is used as the area defined by the contour key information.

[0115] After the quality index value of the time frame image is calculated, it can be detected whether the quality index value of the time frame image meets the preset quality condition. Specifically, the preset quality condition can be a pre-set condition. The preset quality condition can be different according to specific needs. For example, it can be set that the quality index value of the time frame image needs to be greater than a preset threshold; when it is necessary to select an image as a key frame from multiple images, the preset quality condition can also be set that the quality index value of the time frame image is the highest among the multiple images; when it is necessary to compare with the current key frame to determine whether to use the time frame image to update the current key frame, the preset quality condition can also be set that the quality index value of the time frame image needs to be greater than the quality index value of the current key frame.

[0116] The key information of the contour includes key point information, and the quality index value includes a contour surface index value and a distance index value.

[0117] The step of respectively calculating the quality index value of each time frame image according to the area defined by each contour key information comprises:

[0118] Step S221, determining the contour surface area formed by each key point in each time frame image and the actual position of the target detection object in each time frame image according to each key point information;

[0119] In the present embodiment, it should be noted that the contour key information may include the key point information of the target detection object. The preset target detection model can be called to identify the time frame image to obtain the key point information of the target detection object in the time frame image. Among them, the preset target detection model can be a pre-set and trained model, and the model structure can adopt the existing commonly used target detection model, and add a key point detection branch in the target detection model, that is, a branch for extracting the key points of the target detection object in the image. A large number of images containing target detection objects can be obtained in advance, and the key points of the target detection objects in the image are marked; the key points can be taken as the key points that most of the target detection objects of this type have, for example, when the target detection object is a rectangular parallelepiped-shaped cargo, the key points can be the 8 corner points of the rectangular parallelepiped cargo, and it should be noted that only 7 of the 8 points of the cargo in the image are visible at the same time, so the marking of the key points of the target detection object in the image can include the position of the key points and whether it is visible; the target detection model is trained by acquiring the image and the marking information of the image, and a target detection model that can identify the key point information of the target detection object in the time frame image is obtained. Then the key point information identified by calling the target detection model may include information such as the position and visibility of the key points of the target detection object in the image.

[0120] According to each of the key point information, the contour surface area formed by each of the key points in each of the time frame images and the actual position of the target detection object in each of the time frame images are determined respectively. Specifically, based on the key point information of the target detection object in the time frame image, the position of the target detection object in the image, that is, the area where it is located, can be determined. In a feasible implementation, the distance and position relationship between each visible key point can be determined according to the position of each key point in the key point information and the information whether it is visible. Based on the distance and position relationship, the key points at the edge of each visible key point are sequentially connected to obtain the area defined by each visible key point in the image, and the contour surface area formed by each key point in the time frame image is determined according to the key point information. In addition, it can be determined whether each key point is visible and the position of the visible key point in the image according to the key point information; after determining the position of the visible key point in the image, the distance and position relationship between each key point can be determined. According to the distance and position relationship, which key points can constitute the contour surface area of ​​the target detection object, and it can be determined that several contour surface areas are included. For example, when the target detection object is a rectangular-shaped cargo, it is visible to identify the seven corner points of the cargo from the time frame image. According to the positions of the seven corner points, the distances and positional relationships between the seven corner points are determined. According to the distances and positional relationships of the seven points, the three contour surface areas of the cargo can be determined, namely, the front view surface, the top view surface and one side view surface. Furthermore, based on the positions of the key points, the actual position of the target detection object in the time frame image is determined. In a feasible implementation scheme, the position of the geometric center point of each key point can be used as the actual position of the target detection object in the time frame image, or a target key point can be selected from each key point as the actual position of the target detection object.

[0121] Step S222, respectively calculating the contour surface index value of each contour surface area under a preset contour surface quality index;

[0122] In this embodiment, the contour surface quality index can be set in advance. For example, the preset contour surface quality index can include at least one or more of the area of ​​the contour surface region, the clarity of the contour surface region, and the visibility of the contour surface region. After the contour surface region is determined, the contour surface index value of the contour surface region in each of the time frame images under the preset contour surface quality index can be calculated. Specifically, when there are multiple contour surface regions in the time frame image, the contour surface index value of each contour surface region is calculated separately, or each contour surface region can be regarded as a whole region to calculate the contour surface index value. For example, the clarity, visibility, and area of ​​each contour surface region can be calculated, and the calculated result is used as the contour surface index value. Among them, the calculation method of clarity, visibility, and area can adopt the existing common calculation method, which will not be described in detail here.

[0123] In a feasible implementation, if multiple contour surfaces of the target detection object in each of the time frame images are determined, and each index value of each contour surface is calculated, then according to a preset image quality calculation formula, each index value of each contour surface of the target detection object in each of the time frame images is substituted into the formula to calculate an image quality value of each of the time frame images; then, the image quality value is respectively detected to see whether it is greater than a preset quality value, and if so, the time frame image is used as a key frame; or the image quality value is respectively detected to see whether it is the image quality value of the current key frame, and if so, the time frame image is used as a new key frame to update the original key frame. Among them, the image quality calculation formula can be set according to different specific application scenarios and specific needs. For example, it can be set to multiply each index value of each contour surface to obtain the quality value of each contour surface, and then add the quality value of each contour surface to obtain the result as the image quality value of the time frame image.

[0124] Step S223, respectively calculating the distance between each of the actual positions and the preset optimal position, and obtaining the distance index value corresponding to each of the time frame images.

[0125] In the present embodiment, it should be noted that, by setting up the camera device in advance, according to the image captured by the camera device, it is determined that when the target detection object is in which position in the image, it is most conducive to the recognition of the target detection object, and the position is set to the optimal position. For example, according to the angle at which the camera device is set up, when the target detection object is in the middle of the image captured by the camera device, it is most conducive to the recognition of the target detection object, and the middle position is determined as the optimal position. After calculating the distance between the actual position of the target detection object in the image and the optimal position, the distance can be used as the distance index value of the time frame image. It can be understood that when the distance index value is smaller, the time frame image is more suitable as a key frame.

[0126] Further, when the quality index value includes a distance index value and a contour surface index value, it can be determined whether the time frame image is suitable as a key frame according to the distance index value and the contour surface index value. Specifically, the preset quality condition can be that when the distance index value is less than a threshold value and the contour surface index value is greater than a threshold value, it is determined that the quality index value meets the preset quality condition, wherein the two threshold values ​​can be set according to specific needs.

[0127] In addition, in a feasible implementation, the quality index value may also only include the distance index value, that is, whether the time frame image is suitable as a key frame can be judged only based on the distance index value of the time frame image. The preset quality condition can be that when the distance index value of the time frame image is less than a threshold, it is determined that the quality index value meets the preset quality condition, or the preset quality condition can be that when the distance index value of the time frame image is smaller than the distance index value of the current key frame, it is determined that the quality index value meets the preset quality condition.

[0128] In this embodiment, the contour surface area formed by each key point is determined according to the key point information of the target detection object in the time frame image, and the contour surface index value of the contour surface area under the preset contour surface quality index is used as the quality index value of the time frame image, so as to achieve the refinement of the image to each contour surface of the target detection object, and determine whether the time frame image is suitable as a key frame according to the index value of the contour surface, so that the finally determined key frame is more appropriate and accurate.

[0129] Further, refer to Figure 3 Based on the first and second embodiments of the present application, in another embodiment of the present application, the quality index value includes a contour surface index value and a distance index value, and each of the time frame images includes a current time frame image.

[0130] The step of determining a key frame in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values ​​comprises:

[0131] Step A10, determining the current key frame of each preset camera, and respectively obtaining the key frame contour surface index value and the key frame distance index value corresponding to each current key frame;

[0132] In this embodiment, the current key frame corresponding to the time frame image can be determined, and the key frame quality value and key frame distance index value corresponding to the current key frame can be obtained. Among them, the current key frame is taken before the time frame image and has been temporarily determined as a key frame for identifying the target detection object, the key frame contour surface index value is the contour surface index value of the current key frame, and the key frame distance index value is the distance index value of the current key frame.

[0133] Wherein, the step of determining the current key frame of each preset camera includes:

[0134] The following steps may be performed for each of the time frame images to determine the current key frame corresponding to the time frame image:

[0135] Step A11, inputting the key point information of the time frame image and each previous key point information in the corresponding previous image into a preset tracking algorithm, and obtaining the corresponding matching degree between the key point information and each previous key point information;

[0136] In the present embodiment, it should be noted that when there are multiple target detection objects in the time frame image, for one of the target detection objects, in order to determine the current key frame corresponding to the target detection object, a previous image of the time frame image can be obtained, and the previous image can be an image taken before the adjacent time frame image in the image of continuous capture by the same preset camera. It is understandable that the previous image has been judged whether it can replace the current key frame. The key point information of the target detection object in the time frame image is respectively input into the preset tracking algorithm with each previous key point information in the previous image, and the matching degree corresponding to the key point information and each previous key point is obtained. Wherein, the preset tracking algorithm can be a tracking algorithm set in advance, for example, sort algorithm (a kind of multi-target tracking algorithm). The principle of the tracking algorithm is to determine whether the target detection object corresponding to two key point information is the same based on the position of the key point in the key point information.

[0137] Step A12, based on the matching degrees, determining the target previous key point information having the greatest matching degree with the key point information;

[0138] In this embodiment, after obtaining the matching degree between the key point information of the target detection object and each previous key point information, the previous key point information with the highest matching degree with the key point information of the target detection object is selected from each previous key point information as the target previous key point information. It should be noted that the target detection object corresponding to the previous key point with the highest matching degree is most likely to be the same target detection object as the target detection object.

[0139] Step A13: taking the current time frame image corresponding to the previous key point information of the target as the current key frame.

[0140] In this embodiment, the current key frame corresponding to the target's previous key point information is used as the current key frame corresponding to the time frame image. That is, the corresponding relationship between the target detection objects in each previous image and the current key frame of each target detection object have been calculated in advance. For the latest captured time frame image, a certain target detection object in the time frame image is determined by a tracking algorithm, and which previous target detection object it corresponds to, so that the current key frame corresponding to the target detection object can be determined, that is, the current key frame of the time frame image is determined.

[0141] Step A20, respectively detecting whether the distance ratio between the distance index value of each current time frame image and the corresponding key frame distance index value is less than a preset ratio, and detecting whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value;

[0142] In this embodiment, the current time frame image is a time frame image captured by the corresponding preset camera in the current time frame, and the distance ratio between the distance index value of each current time frame image and the distance index value of the key frame is detected to see whether it is less than the preset ratio. That is, the ratio of the distance index value of the time frame image to the distance index value of the key frame, and the ratio is detected to see whether it is less than the preset ratio. Among them, the preset ratio can be set as needed, for example, it is set to 1.2, that is, the distance index value of the time frame image cannot exceed 1.2 times the distance index value of the key frame. And the contour surface index value of each time frame image is detected to see whether it is greater than the contour surface index value of the key frame.

[0143] Step A30, if the distance ratio is less than the preset ratio, and the contour surface index value is greater than the key frame contour surface index value, the current time frame image is updated to the current key frame until all the time frame images are detected and all the key frames are obtained.

[0144] In the present embodiment, if the distance ratio is less than the preset ratio, and the profile surface index value is greater than the key frame profile surface index value, then the time frame image can be used as a key frame for identifying the target detection object to replace the current key frame to complete the update of the key frame. The updated key frame is more suitable for identifying the target detection object than the key frame before the update, and then the accuracy of the key frame selection can be improved, and the accuracy of the target detection object identification detection can be improved. If it is detected that the distance ratio is not less than the preset ratio, or the profile surface index value is detected to be not greater than the key frame profile surface index value, the current key frame will still be retained, and then after each time frame image captured by each of the preset cameras is detected, the key frame corresponding to each of the preset cameras is determined.

[0145] That is, each of the preset cameras can continuously take multiple images of the target object to be identified, and process each time frame image in sequence according to the time sequence of the shooting; for the first time frame image, the time frame image can be used as a key frame, and for subsequent time frame images, if the contour surface index value is larger than that of the current key frame and the distance index value does not exceed the preset multiple of the distance index value of the current key frame, the time frame image can replace the current key frame, otherwise the current key frame is retained, and then after the time frame images corresponding to each of the preset cameras are detected according to the time sequence of the corresponding time frames, the key frames corresponding to each camera are obtained.

[0146] This embodiment is based on determining the current key frame of each of the preset cameras, and respectively obtaining the key frame contour surface index value and the key frame distance index value corresponding to each of the current key frames, and then respectively detecting whether the distance ratio between the distance index value of each of the current time frame images and the corresponding key frame distance index value is less than the preset ratio, and detecting whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value, and then if the distance ratio is less than the preset ratio and the contour surface index value is greater than the key frame contour surface index value, the current time frame image is updated to the current key frame, until all the time frame images are detected and each of the key frames is obtained. That is, this embodiment provides a key frame selection method, that is, the time frame image corresponding to the target detection object is screened from the contour surface of the target detection object and the actual position of the target detection object in the image at the same time, and the key frame finally obtained meets both the contour surface index and the distance index value, thereby achieving the purpose of evaluating the image quality of the time frame image from multiple aspects, making the final key frame more conducive to the identification of the target detection object, providing a stable and feasible solution for the selection of key frames, and further laying the foundation for improving the accuracy of object recognition detection.

[0147] Reference Figure 4 , Figure 4 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present application.

[0148] like Figure 4 As shown, the object identification and detection device may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the processor 1001 and the memory 1005. The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0149] Optionally, the object recognition and detection device may also include a rectangular user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, etc. The rectangular user interface may include a display screen (Display), an input submodule such as a keyboard (Keyboard), and the optional rectangular user interface may also include a standard wired interface and a wireless interface. The network interface may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0150] Those skilled in the art will understand that Figure 4The structure of the object identification and detection device shown in the figure does not constitute a limitation on the object identification and detection device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0151] like Figure 4 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, and an object identification and detection method program. The operating system is a program that manages and controls the hardware and software resources of the object identification and detection device, and supports the operation of the object identification and detection method program and other software and / or programs. The network communication module is used to realize the communication between the components inside the memory 1005, and the communication with other hardware and software in the object identification and detection method system.

[0152] exist Figure 4 In the object recognition and detection device shown, the processor 1001 is used to execute the object recognition and detection method program stored in the memory 1005 to implement the steps of any of the above-mentioned object recognition and detection methods.

[0153] The specific implementation of the object identification and detection device of the present application is basically the same as the various embodiments of the above-mentioned object identification and detection method, and will not be repeated here.

[0154] The present application also provides an object recognition and detection device, the object recognition and detection device comprising:

[0155] An acquisition module, used to acquire time frame images taken by each preset camera in each time frame, and determine the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera;

[0156] A first determination module is used to determine the target detection object corresponding to the object correspondence relationship, and calculate the quality index value of each of the time frame images for the target detection object;

[0157] A second determination module is used to determine key frames in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values;

[0158] The recognition and detection module is used to recognize and detect the target detection object based on each of the key frames to obtain a target recognition and detection result.

[0159] Optionally, the acquisition module includes:

[0160] A selection unit, configured to select a reference camera from among the preset cameras and determine a reference coordinate system corresponding to the reference camera;

[0161] A conversion unit, configured to convert the to-be-determined detection object in each of the time frame images into the reference coordinate system based on the coordinate conversion relationship, and calculate similarity index values ​​between each of the to-be-determined detection objects in the reference coordinate system;

[0162] The first determining unit is used to determine the object correspondence based on each of the similarity index values.

[0163] Optionally, the conversion unit includes:

[0164] A first determination subunit is used to determine the intersection part and the union part between each of the to-be-determined detection objects in the reference coordinate system;

[0165] The first calculation subunit is used to calculate each of the similarity index values ​​based on each of the intersection parts and each of the union parts.

[0166] Optionally, the identification and detection module includes:

[0167] An identification and detection unit, used to input each of the key frames into a preset object identification and detection model, respectively, to identify and detect the target detection object in each of the key frames, and obtain an identification and detection result corresponding to each of the key frames;

[0168] The second determination unit is used to determine the target recognition detection result in each of the recognition detection results based on the key frame confidence corresponding to each of the key frames.

[0169] Optionally, the first determining module includes:

[0170] An identification unit, used to identify each of the time frame images respectively, and obtain key information of the contour of the target detection object in each of the time frame images;

[0171] The calculation unit is used to calculate the quality index value of each time frame image according to the area defined by each contour key information.

[0172] Optionally, the computing unit includes:

[0173] A second determination subunit is used to determine the contour surface area formed by each key point in each time frame image and the actual position of the target detection object in each time frame image according to each key point information;

[0174] A second calculation subunit, used to respectively calculate the contour surface index value of each contour surface area under a preset contour surface quality index;

[0175] The third calculation subunit is used to respectively calculate the distance between each of the actual positions and the preset optimal position, and obtain the distance index value corresponding to each of the time frame images.

[0176] Optionally, the second determining module includes:

[0177] An acquisition unit, used to determine the current key frame of each preset camera, and respectively acquire the key frame contour surface index value and the key frame distance index value corresponding to each current key frame;

[0178] A detection unit, used to respectively detect whether the distance ratio between the distance index value of each current time frame image and the corresponding key frame distance index value is less than a preset ratio, and to detect whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value;

[0179] An updating unit is used to update the current time frame image to the current key frame if the distance ratio is less than the preset ratio and the contour surface index value is greater than the key frame contour surface index value, until all the time frame images are detected and all the key frames are obtained.

[0180] The specific implementation of the object identification and detection device of the present application is basically the same as the various embodiments of the above-mentioned object identification and detection method, and will not be repeated here.

[0181] An embodiment of the present application provides a readable storage medium, and the readable storage medium stores one or more programs, and the one or more programs can also be executed by one or more processors to implement the steps of any of the above-mentioned object identification and detection methods.

[0182] The specific implementation of the readable storage medium of the present application is basically the same as the embodiments of the above-mentioned object identification and detection method, and will not be repeated here.

[0183] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A method for object identification and detection, characterized in that: The object identification and detection method comprises: Acquire time frame images taken by each preset camera in each time frame, and determine the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera, wherein each preset camera is arranged at different shooting positions and shooting angles; Determine the target detection object corresponding to the object correspondence relationship, and calculate the quality index value of each of the time frame images for the target detection object; Based on the quality index values, key frames are determined in the time frame images corresponding to the preset cameras respectively; Input each of the key frames into a preset object recognition and detection model, respectively, perform recognition and detection on the target detection object in each of the key frames, and obtain recognition and detection results corresponding to each of the key frames; Based on the key frame confidences corresponding to the key frames, selecting the recognition detection result with the highest key frame confidence as the target recognition detection result; Wherein, the transformation relationship includes a coordinate conversion relationship, and the step of determining the object correspondence relationship between the time frame images corresponding to each of the time frames based on the transformation relationship between the preset cameras includes: Selecting a reference camera from each of the preset cameras, and determining a reference coordinate system corresponding to the reference camera; Based on the coordinate transformation relationship, the to-be-determined detection object in each of the time frame images is transformed into the reference coordinate system, and the similarity index values ​​between each of the to-be-determined detection objects are calculated in the reference coordinate system, wherein the coordinate transformation relationship is a transformation relationship between the plane coordinate systems of the shooting ground corresponding to each of the preset cameras; Determining the object correspondence based on the similarity index values; Wherein, the quality index value includes a contour surface index value and a distance index value, each of the time frame images includes a current time frame image, The step of determining a key frame in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values ​​comprises: Determine the current key frame of each preset camera, and respectively obtain the key frame contour surface index value and the key frame distance index value corresponding to each current key frame; Respectively detecting whether the distance ratio between the distance index value of each current time frame image and the corresponding key frame distance index value is less than a preset ratio, and detecting whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value; If the distance ratio is less than the preset ratio and the contour surface index value is greater than the key frame contour surface index value, the current time frame image is updated to the current key frame until all the time frame images are detected and each key frame is obtained.

2. The object identification and detection method according to claim 1, characterized in that: The step of calculating the similarity index values ​​between each pair of the to-be-determined detection objects in the reference coordinate system comprises: Determine the intersection and union of each of the to-be-determined detection objects in the reference coordinate system; Based on the intersection parts and the union parts, the similarity index values ​​are calculated.

3. The object identification and detection method according to claim 1, characterized in that: The step of calculating the quality index value of each of the time frame images for the target detection object comprises: Respectively identifying each of the time frame images to obtain key information of the contour of the target detection object in each of the time frame images; According to the area defined by each of the contour key information, the quality index value of each of the time frame images is calculated respectively.

4. The object identification and detection method according to claim 3, characterized in that: The contour key information includes key point information, and the quality index value includes a contour surface index value and a distance index value. The step of respectively calculating the quality index value of each time frame image according to the area defined by each contour key information comprises: According to the key point information, respectively determine the contour surface area formed by each key point in each time frame image and the actual position of the target detection object in each time frame image; Respectively calculating the profile surface index value of each profile surface area under a preset profile surface quality index; The distance between each of the actual positions and the preset optimal position is calculated respectively to obtain the distance index value corresponding to each of the time frame images.

5. An object identification and detection device, characterized in that: The object identification and detection device comprises: An acquisition module is used to acquire time frame images taken by each preset camera in each time frame, and determine the object correspondence between each time frame image corresponding to each time frame based on the transformation relationship between each preset camera, wherein each preset camera is arranged at different shooting positions and shooting angles; A first determination module is used to determine the target detection object corresponding to the object correspondence relationship, and calculate the quality index value of each of the time frame images for the target detection object; A second determination module is used to determine key frames in each of the time frame images corresponding to each of the preset cameras based on each of the quality indicator values; The recognition and detection module is used to input each of the key frames into a preset object recognition and detection model, respectively, to recognize and detect the target detection object in each of the key frames, respectively, to obtain the recognition and detection results corresponding to each of the key frames, and based on the key frame confidences corresponding to each of the key frames, to select the recognition and detection result with the highest key frame confidence as the target recognition and detection result; Wherein, the transformation relationship includes a coordinate conversion relationship, and the acquisition module is further used to: select a reference camera from each of the preset cameras, and determine a reference coordinate system corresponding to the reference camera; based on the coordinate conversion relationship, convert the to-be-determined detection objects in each of the time frame images into the reference coordinate system, and calculate the similarity index values ​​between each of the to-be-determined detection objects in the reference coordinate system, wherein the coordinate conversion relationship is a conversion relationship between the plane coordinate systems between the shooting grounds corresponding to each of the preset cameras; based on each of the similarity index values, determine the object correspondence; Among them, the quality index value includes a contour surface index value and a distance index value, each of the time frame images includes a current time frame image, and the second determination module is also used to: determine the current key frame of each of the preset cameras, and respectively obtain the key frame contour surface index value and key frame distance index value corresponding to each of the current key frames; respectively detect whether the distance ratio between the distance index value of each of the current time frame images and the corresponding key frame distance index value is less than the preset ratio, and detect whether the contour surface index value of the current time frame image is greater than the corresponding key frame contour surface index value; if the distance ratio is less than the preset ratio, and the contour surface index value is greater than the key frame contour surface index value, then update the current time frame image to the current key frame, until all the time frame images are detected and each of the key frames is obtained.

6. An object identification and detection device, characterized in that: The object recognition and detection device comprises: a memory, a processor, and a program stored in the memory for implementing the object recognition and detection method. The memory is used to store a program for implementing the object identification and detection method; The processor is used to execute a program for implementing the object recognition and detection method, so as to implement the steps of the object recognition and detection method as claimed in any one of claims 1 to 4.

7. A readable storage medium, characterized in that: The readable storage medium stores a program for implementing the object identification and detection method, and the program for implementing the object identification and detection method is executed by a processor to implement the steps of the object identification and detection method as claimed in any one of claims 1 to 4.

Images