CN115620047A

CN115620047A - Target object attribute information determination method and device, electronic equipment and storage medium

Info

Publication number: CN115620047A
Application number: CN202211183267.XA
Authority: CN
Inventors: 周勋
Original assignee: China Automotive Innovation Co Ltd
Current assignee: China Automotive Innovation Co Ltd
Priority date: 2022-09-27
Filing date: 2022-09-27
Publication date: 2023-01-17

Abstract

The application relates to a method and a device for determining attribute information of a target object, electronic equipment and a storage medium, wherein the method comprises the following steps: carrying out target object detection on the image of the surrounding environment of the vehicle to obtain a target object detection result; when the target object detection result indicates that the target object exists, acquiring position information of the target object in the image to be detected and first attribute information of the target object from the target object detection result; the first attribute information represents the type of the target object; the target object is an object that has an impact on intelligent driving decisions; determining a region to be processed corresponding to the position information from the image to be detected; when the first attribute information indicates that the target object is of the first type, performing second attribute extraction on the region to be processed to obtain second attribute information of the target object; the second attribute information is used to determine a subdivision type of the target object. Therefore, the method and the device can improve the comprehensiveness and accuracy of target object information detection and are beneficial to improving the reasonability of intelligent driving decision.

Description

Target object attribute information determination method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of image detection technologies, and in particular, to a method and an apparatus for determining attribute information of a target object, an electronic device, and a storage medium.

Background

At present, image detection technology is widely applied in different fields, for example, in the field of automatic driving, a vehicle can be ensured to run safely by detecting a target object such as an obstacle, a pedestrian, road facilities and the like in an acquired image of the surrounding environment of the vehicle and transmitting a detection result to an automatic driving decision module.

In the related art, most of the automatic driving decisions are made according to the types of the target objects by identifying the types of the target objects, but the type identification of the target objects depends too much on the classification capability of a model or an algorithm, and the target objects with more types and complicated types can be classified only in coarse granularity, so that the target objects are not known enough, and effective information which is helpful for the automatic driving decisions cannot be provided.

Disclosure of Invention

The application provides a method and a device for determining attribute information of a target object, electronic equipment and a storage medium, and the technical scheme of the application is as follows:

according to a first aspect of the embodiments of the present application, a method for determining attribute information of a target object is provided, including:

carrying out target object detection on an image to be detected to obtain a target object detection result; the image to be detected is an image of the surrounding environment of the vehicle;

when the target object detection result indicates that the target object exists, acquiring position information of the target object in the image to be detected and first attribute information of the target object from the target object detection result; the first attribute information represents the type of the target object; the target object is an object that has an impact on intelligent driving decisions;

determining a region to be processed corresponding to the position information from the image to be detected;

when the first attribute information indicates that the target object is of the first type, performing second attribute extraction on the region to be processed to obtain second attribute information of the target object; the second attribute information is used for determining a first subtype of the target object; the first subtype is a first type of subdivision type.

In some possible embodiments, the method further comprises:

when the first attribute information indicates that the target object is of the second type, performing third attribute extraction on the region to be processed to obtain third attribute information of the target object; the third attribute information is supplementary information of a second type.

In some possible embodiments, the detecting a target object on an image to be detected to obtain a target object detection result includes:

carrying out target object detection on an image to be detected according to the target object detection model to obtain a target object detection result;

when the target object detection result indicates that the target object exists, the target object detection result comprises coordinate information and size information of a target object detection frame; the coordinate information and the size information are used for determining the position information of the target object in the image to be detected.

In some possible embodiments, the generating manner of the target object detection model includes:

acquiring a labeled first training image set; each frame of the first training image in the first training image set comprises a target object;

constructing a preset machine learning model;

performing target object detection training on a preset machine learning model according to the first training image set until an initial training end condition is met to obtain an initial target object detection model;

automatically labeling each frame of second training image in the unlabeled second training image set according to the initial target object detection model to obtain a labeled second training image set;

verifying the labeled second training image set to obtain a verified second training image set;

and training the initial target object model according to the verified second training image set until a preset training end condition is met, and obtaining a target object detection model.

In some possible embodiments, the second attribute information includes dominant color information;

performing second attribute extraction on the region to be processed to obtain second attribute information of the target object, wherein the second attribute information comprises:

and extracting the main color of the area to be processed to obtain the main color information of the target object.

In some possible embodiments, the performing dominant color extraction on the region to be processed to obtain dominant color information of the target object includes:

determining a sub-region of interest of each candidate color in the region to be processed;

performing edge extraction on the interesting sub-region corresponding to each candidate color to obtain edge information of each candidate color;

determining a coverage area of each candidate color based on the edge information of each candidate color;

and determining a main color from the candidate colors based on the coverage area of each candidate color to obtain the main color information of the target object.

In some possible embodiments, determining the dominant color from the plurality of candidate colors based on the coverage area of each candidate color, and obtaining dominant color information of the target object includes:

sequencing the candidate colors based on the coverage area of each candidate color to obtain a sequencing result;

if the first candidate color in the sorting result is the first preset color, the second candidate color in the sorting result is obtained;

if the second candidate color is the second preset color, respectively determining the size of a first external rectangle corresponding to the first preset color and the size of a second external rectangle corresponding to the second preset color;

and if the size of the first external rectangle and the size of the second external rectangle meet the first preset condition, taking the second preset color as the main color to obtain the main color information of the target object.

In some possible embodiments, the third attribute information includes character information;

performing third attribute extraction on the region to be processed to obtain third attribute information of the target object, wherein the third attribute information comprises:

and extracting characters from the area to be processed to obtain character information of the target object.

According to a second aspect of the embodiments of the present application, there is provided a target object attribute information determination apparatus, including:

the detection module is used for detecting a target object of the image to be detected to obtain a target object detection result; the image to be detected is an image of the surrounding environment of the vehicle;

the acquisition module is used for acquiring the position information of the target object in the image to be detected and the first attribute information of the target object from the target object detection result when the target object detection result indicates that the target object exists; the first attribute information represents the type of the target object; the target object is an object that has an impact on intelligent driving decisions;

the determining module is used for determining a to-be-processed area corresponding to the position information from the to-be-detected image;

the extraction module is used for performing second attribute extraction on the region to be processed when the first attribute information comprises first type information to obtain second attribute information of the target object; the second attribute information is used for determining first subtype information of the target object; the first subtype corresponding to the first subtype information is a first type of subdivision type corresponding to the first type information.

According to a third aspect of embodiments of the present application, there is provided an electronic apparatus, including:

a processor;

a memory for storing processor-executable instructions;

the processor is configured to execute the instructions to implement the method for determining the attribute information of the target object in the first aspect of the embodiment of the present application.

According to a fourth aspect of embodiments of the present application, there is provided a computer-readable storage medium, where instructions of the computer-readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the target object attribute information determination method of the first aspect of embodiments of the present application.

The technical scheme provided by the embodiment of the application at least has the following beneficial effects:

the method comprises the steps of detecting a target object by carrying out target object detection on an image of the surrounding environment of a vehicle to obtain first attribute information, namely type, of the target object, and when the target object is of the first type, extracting second attribute of an area where the target object is located to obtain second attribute information of the target object, so that the subdivision type of the target object can be further determined based on the second attribute information, and the problem that the subdivision type of the target object cannot be directly obtained from the image in the related technology can be solved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.

Drawings

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 application and are not to be construed as limiting the application.

FIG. 1 is a schematic diagram of an application environment shown in accordance with an exemplary embodiment;

FIG. 2 is a flow diagram illustrating a method for determining target object attribute information in accordance with an exemplary embodiment;

FIG. 3 is a schematic diagram illustrating a target object detection process in accordance with an exemplary embodiment;

FIG. 4 is a flowchart illustrating a manner of generating a target object detection model in accordance with an exemplary embodiment;

FIG. 5 is a flowchart illustrating dominant color extraction for a region to be processed in accordance with an illustrative embodiment;

FIG. 6 is a flow diagram illustrating a verification of dominant color information of a target object in accordance with an exemplary embodiment;

FIG. 7 is a block diagram illustrating a target object attribute information determination apparatus in accordance with an illustrative embodiment;

FIG. 8 is a block diagram illustrating an electronic device for target object attribute information determination, according to an example embodiment.

Detailed Description

In order to make the technical solutions of the present application better understood by those of ordinary skill in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in sequences other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

In recent years, automatic driving has been rapidly developed. Automatic driving is one of the most topical fields in the automobile industry in recent years, and the top scientific and technological companies and automobile enterprises all over the world are almost invested in the wave of research and development of automatic driving automobiles. Especially, the popularization of intelligent driving assistance in recent years provides great convenience for consumers.

Based on this, the embodiment of the application provides a method for determining attribute information of a target object, which can be used in an intelligent driving application scene, and can determine multiple kinds of attribute information of the target object by detecting the target object in a collected image so as to assist an automatic driving vehicle to understand the target object, wherein the target object is an object influencing an intelligent driving decision, so that the driving safety of the automatic driving vehicle can be improved. The target objects may include obstacles, pedestrians, road facilities (such as traffic signs and traffic lights), etc. according to different practical application environments.

For example, for an autonomous automobile, it must be possible to identify and understand traffic signs to ensure that they comply with road regulations. Thus, in a specific application scenario, referring to fig. 1, fig. 1 is a schematic diagram of an application environment according to an exemplary embodiment of the present application, where the application environment includes a vehicle 101, and the target object 102 may be a traffic sign.

As shown in fig. 1, a vehicle 101 continuously acquires an environmental image around the vehicle during driving; then, taking the environment image as an image to be detected, and detecting the traffic sign board on the image to be detected to obtain attribute information of the traffic sign, wherein the attribute information comprises type information, so that corresponding help is uninterruptedly provided for the control of the whole vehicle; for example, when the prohibition type mark is detected, the vehicle can be helped to carry out danger prejudgment in advance; when the indication mark is detected, the vehicle can be assisted in control preprocessing to ensure that the driving follows the road indication.

In some possible embodiments, the vehicle 101 may detect the image to be detected through its own intelligent driving assistance system, and obtain the target object attribute information.

Or, in other possible embodiments, the vehicle 101 may also upload the acquired image to be detected to a server, and the server detects the target object of the image to be detected to obtain the attribute information of the target object.

The service end may include an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, and a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a CDN (Content Delivery Network), a big data and artificial intelligence platform, and the like. The operating system running on the server may include, but is not limited to, an android system, an IOS system, linux, windows, unix, and the like.

In addition, it should be noted that fig. 1 shows only one application environment of the target object attribute information determination method provided in the present application, and in practical applications, other application environments may also be included.

Fig. 2 is a flowchart illustrating a target object attribute information determination method according to an exemplary embodiment, and as shown in fig. 2, the target object attribute information determination may be applied to an intelligent driving assistance system for a vehicle, including the steps of:

in step S201, a target object detection is performed on the image to be detected, so as to obtain a target object detection result.

In step S203, when the target object detection result indicates that the target object exists, acquiring position information of the target object in the image to be detected and first attribute information of the target object from the target object detection result; the first attribute information characterizes a type of the target object.

In the embodiment of the application, wait to examine the image and can be that the car end is gathered, intelligent driving auxiliary system can include on-vehicle camera, acquires the environment image back around the vehicle through on-vehicle camera, regards as waiting to examine the image with the environment image. Or the image to be detected may be obtained from a road environment image library stored in the vehicle, the road environment image library stores a plurality of road environment images with a plurality of pieces of position information corresponding to one another, the intelligent driving assistance system may obtain the road environment image corresponding to the current vehicle positioning information from the road environment image library according to the current vehicle positioning information, and take the road environment image corresponding to the current vehicle positioning information as the image to be detected; or the image to be detected can come from the road end equipment, when the road end equipment senses that the vehicle enters the jurisdiction area of the vehicle, the road end equipment establishes communication connection with the vehicle and sends the currently stored road environment image in the jurisdiction area of the vehicle, and the intelligent driving auxiliary system receives the road environment image and takes the road environment image as the image to be detected.

In the embodiment of the present application, the target object is an object matched with an actual application field and a scene, for example, in the field of automatic driving, the target object is an object having an influence on an intelligent driving decision, and specifically may include an obstacle, a pedestrian, road facilities (such as a traffic sign, a traffic light), and the like. The first attribute information represents the type of the target object, different first attribute information represents different types of the target object, and the type of the target object comprises a subdivision type and/or a rough subdivision type; the specific type is not limited in the application, and the type can be set by combining with an actual application scene.

For example, when the target object is a traffic sign, the type of the target object may include a warning sign class (a sign for warning a vehicle or a pedestrian of a dangerous spot), a prohibition sign class (a sign for prohibiting or restricting a traffic behavior of a vehicle or a pedestrian), an indication sign (a sign for indicating a travel of a vehicle or a pedestrian), and the like; and further, the intelligent parking lot can further comprise a sign board in rainy and snowy days, a highest speed limit sign board, a lowest speed limit sign board, a speed limit releasing sign board, a stop prohibition sign board and the like.

In some possible embodiments, the images to be detected may be acquired in real time based on the current environment or may be acquired in advance and stored in the corresponding device according to the timeliness of the existence of the target object or the object actually characterized by the target object. For example, when the target object includes an obstacle or a pedestrian, which is not an object existing in the current environment for a long time, the image to be detected is an image acquired in real time by the vehicle-mounted camera; for another example, when the target object includes an object fixed in the current environment for a long time, such as a traffic sign, the image to be detected may be a road environment image stored in advance in a road environment image library of the vehicle, or may be a road environment image stored in a road end device.

In the embodiment of the application, after the intelligent driving assistance system obtains the image to be detected, the image to be detected is subjected to target object detection to obtain a target object detection result; the target object detection result indicates whether a target object exists, and when the target object exists, the target object detection result should include position information of the target object in the image to be detected and first attribute information of the target object; wherein the first attribute information characterizes a type of the target object.

In some possible embodiments, the performing target object detection on the image to be detected to obtain a target object detection result may include the following steps: and carrying out target object detection on the image to be detected according to the target object detection model to obtain a target object detection result.

Specifically, as shown in fig. 3, an image to be detected is input into a trained target object detection model, and target object detection is performed on the image to be detected through the model, so as to predict the position and type of the target object; when the target object detection result output by the model comprises the coordinate information and the size information of the target object detection frame, indicating that the target object exists in the image to be detected, and directly determining the coordinate information and the size information { x, y, w, h } of the target object detection frame as the position information of the target object in the image to be detected, wherein x and y represent the central coordinate information of the target object detection frame; w, h respectively represent the width and height of the detection frame, i.e. the size information; meanwhile, the type of the target object detection frame predicted by the model is used as the first attribute information of the target object, in this example, the target object is a traffic sign, and the type of the target object detection frame, that is, the first attribute information is the highest speed limit type.

The accuracy of the target object detection model greatly depends on the data volume during training. In the traditional training process, a training image with labels needs to be obtained in a manual labeling mode. For example, when the image in fig. 3 is labeled, the traffic sign is labeled manually by using a rectangular frame or other polygonal frames, and a corresponding type label is marked, which is a highest speed limit type; however, because the types of the traffic signs in the actual scene are many and are not easy to distinguish, the traditional method not only needs to consume a large amount of manpower and time cost, but also is easy to make mistakes.

Based on this, in this application embodiment, at first train the model based on the training image of artifical mark, obtain the preliminary detection model that possesses certain precision, recycle preliminary detection model and carry out the target object to the image that does not mark and detect to realize the automatic mark of training image, so, through the mode that artifical mark and automatic mark combined together, can accelerate the efficiency of model training, and reduce the error rate of artifical mark.

In some possible embodiments, the generation manner of the target object detection model may include the following steps as shown in fig. 4:

in step S401, a labeled first training image set is obtained; each frame of the first training image in the first set of training images includes a target object.

Wherein, the labeled first training image set can be directly obtained from the existing training data set in the related field or the related application scene; for example, when the method is applied to detecting a Traffic Sign, the labeled first training image set may be directly obtained from a data set such as a Chinese Traffic Sign Detection dataset (CSUST Chinese Traffic Sign Detection Benchmark, CCTSDB), and a TT-100K data set.

Or, the labeled first training image set may also be acquired by an image acquisition vehicle or other acquisition methods, and then the acquired images are manually labeled. Therefore, a training data set matched with the actual application scene can be established according to the actual application scene, and the problems that labeled information in the existing training data set is not adaptive to the actual application scene, such as different types of classification modes, can be solved.

Because the first training image set is labeled, each frame of the first training image set carries labeling information, and the labeling information includes coordinate information and a type of a real frame where the target object is located.

In step S403, a preset machine learning model is constructed.

The preset machine learning model can adopt the existing target detection algorithm model structure in the computer vision related field as a basic model; such as regional convolutional neural networks (R-CNN), fast R-CNN, YOLO algorithm models, and the like. In practical application, the basic model can be adjusted or improved to adapt to practical requirements.

In a specific example, the preset machine learning model may adopt the structure of the YOLO v5 model.

In step S405, performing target object detection training on a preset machine learning model according to the first training image set until an initial training end condition is satisfied, so as to obtain an initial target object detection model.

Specifically, each frame of first training image in a first training image set is input into a preset machine learning model, the position and the type of a target object in the first training image are predicted, a target object prediction frame is output, coordinate information of the target object prediction frame is compared with coordinate information of a labeled real frame, the type of the identified target object prediction frame is compared with the labeled type, and a loss value is determined based on a coordinate information comparison result and a type comparison result; then, performing back propagation on the basis of the loss value, and performing iterative updating on the model to obtain an updated model; then, repeating the above steps; obtaining an initial target object detection model until an initial training ending condition is met; the initial training end condition may be that training is ended when the iteration number reaches a first preset number, or may also be that training is ended when the loss value gradually converges to a first threshold; here, the first preset number of times may be 50 times, 100 times, etc. and the first threshold value may be 0.2, 0.4, etc. by way of example.

In step S407, each frame of second training image in the unlabeled second training image set is automatically labeled according to the initial target object detection model, so as to obtain a labeled second training image set.

In the step, an initial target object detection model which is obtained after preliminary training and has certain precision is used for automatically labeling a second unlabeled training image set to obtain a labeled second training image set; and each frame of second training image in the labeled second training image set carries model automatic labeling information. The second training image set which is not marked is similar to the first training image set, can be directly obtained from the existing training data set in the related field or the related application scene, and can also be obtained by collecting through an image collecting vehicle or other collecting modes.

Specifically, each frame of second training image in the unlabelled second training image set is input into an initial target object detection model, target object detection is performed on each frame of second training image through the initial target object detection model, and an initial target object detection result of each frame of second training image is output; obtaining model automatic labeling information of each frame of second training image based on an initial target object detection result of each frame of second training image; when the second training image does not contain the target object, the corresponding initial target object detection result indicates that the target object does not exist, and the model automatic labeling information indicates that the image is an invalid training image; and when the second training image contains the target object, and the corresponding initial target object detection result comprises the initial target object prediction frame and the initial type of the target object, taking the initial target object prediction frame and the initial type of the target object as model automatic labeling information.

In step S409, the labeled second training image set is verified to obtain a verified second training image set.

In this step, after obtaining the labeled second training image set based on the initial target object detection model, the labeled second training image set may be verified based on the aforementioned model automatic labeling information. Specifically, at least one of the following verification methods may be adopted:

(1) Deleting the second training image of which the model automatic labeling information is indicated as an invalid training image;

(2) Manually judging whether the types of the target object prediction frame and the target object in the target object detection result are accurate or not, and deleting or correcting a second training image with the inaccurate types of the target object prediction frame and the target object;

in this way, the labeled second training image set is verified through the at least one verification mode, so that a verified second training image set is obtained.

In step S411, the initial target object model is trained according to the verified second training image set until a preset training end condition is met, so as to obtain a target object detection model.

In the step, the verified second training image set is used again to train the initial target object model; inputting each frame of second training image in the verified second training image set into the initial target object model again, and performing target object detection on each frame of second training image through the initial target object model again to obtain a current prediction frame of the target object and a current type of the target object; then comparing the current prediction frame of the target object with the initial prediction frame (passed by the verification) of the target object, comparing the initial type of the target object with the current type of the target object, determining the loss value again based on the comparison result of the prediction frames and the comparison result of the types, performing back propagation by using the loss value, and performing iterative update on the initial target object detection model to obtain an updated model; then, repeating the above steps; and obtaining the trained target object detection model until the preset training end condition is met. The preset training ending condition may be that the training is ended when the iteration number reaches a second preset number, or may also be that the training is ended when the loss value gradually converges to a second threshold value; here, as an example, the second preset number of times may be 50 times, 100 times, etc., and the second threshold may be 0.1; in general, the second threshold is smaller than the first threshold, because the target object detection model is different from the initial target object detection model, the initial target object detection model is mainly used for automatic annotation of a training image, and the target object detection model finally needs to detect a target object in an actual application scene.

In the embodiment, the model is preliminarily trained on the basis of the manually marked training image to obtain an initial target object detection model with certain precision, and the automatic marking of the training image is realized through the initial target object detection model, so that the manual marking time of marking personnel can be saved, and the error rate of manual marking is reduced; after the automatic marking is finished, manual verification is carried out, so that the integrity of the automatic marking information of the model and the effectiveness of the training image can be ensured; and finally, finishing the training of the final target object detection model based on the automatically labeled training image.

It should be noted that, in other possible embodiments of the present application, the step S411 may be replaced by the following step S413:

in step S413, the newly constructed machine learning model is trained by using the first training image set and the verified second training image set until a preset training end condition is met, so as to obtain a target object detection model.

Here, the initial target object detection model is replaced by constructing a new machine learning model; training on the basis of a newly constructed machine learning model by utilizing a first training image set labeled manually and a second training image set labeled automatically to obtain a target object detection model; the method has the advantages that the automatic labeling function and the target object detection function are decoupled, two models which independently realize different functions are obtained, different models can be called according to different requirements in actual application, and flexibility is high.

In the related art, after the position and the type of a target object are determined from an image to be detected, an intelligent driving assistance system immediately executes an automatic driving decision according to the position and the type of the target object. However, in most scenarios, the intelligent driving assistance system can only predict the approximate type of the target object, or classify the target object into a rough classification type for the types that cannot be subdivided, which results in that the intelligent driving assistance system cannot understand the target object well, so that the automatic driving decision made based on the approximate type of the target object only is not ideal in reliability and safety. On the other hand, if the subdivision type of the target object is to be directly detected based on the image, a large amount of image data needs to be collected for each subdivision type at the time of annotation, which imposes a high demand on both data collection and annotation.

Based on this, through the following steps S205 to S209, after the position information of the target object in the image to be detected is determined, the intelligent driving assistance system further extracts other attributes of the region to be processed corresponding to the position information in the image to be detected, so as to obtain other attribute information of the target object except for the type; the target object is further understood through other attribute information, and the intelligent driving assistance system is facilitated to make safer and more reliable decisions.

In a specific example, when the target object is a traffic sign, the traffic sign has various attributes such as colors, characters, shapes and the like besides different types, and the attributes can be used for automatic driving decision, so that the reasonability, the safety and the reliability of the automatic driving decision are improved; even on non-autonomous vehicles, intelligent driving assistance systems can provide information about road conditions and reasonable advice to drivers to assist them in complying with current road regulations.

In step S205, a region to be processed corresponding to the position information is determined from the image to be detected.

In the embodiment of the application, in order to facilitate the determination of other attribute information of the target object except for the type, after the position of the target object in the image to be detected is determined, the intelligent driving assistance system can remove the part, which is irrelevant to the target object, in the image to be detected; that is, the intelligent driving assistance system determines a region to be processed containing the target object from the image to be detected based on the position information of the target object; subsequently, the intelligent driving assistance system mainly extracts the attributes of the target object in the area to be processed to obtain other attribute information of the target object except for the type.

In some possible embodiments, as described above, the position information of the target object includes the coordinate information and the size information { x, y, w, h } of the target object detection frame, so that the intelligent driving assistance system may determine a rectangular region to be processed based on the coordinate information and the size information { x, y, w, h } of the target object detection frame. Thus, the sizes of the respective regions to be processed may be different for different target objects or different types of target objects.

Alternatively, in other possible embodiments, in order to facilitate the extraction of other attributes subsequently, the size of the region to be processed is preset to be a fixed size, and the specific value of the fixed size may be set according to an actual empirical value. The intelligent driving assistance system can use the coordinate information of the target object detection frame as area center coordinate information, and then determines an area to be processed from the image to be detected based on the area center coordinate information and a preset fixed size.

In step S207, when the first attribute information indicates that the target object is of the first type, performing second attribute extraction on the region to be processed to obtain second attribute information of the target object; the second attribute information is used for determining a first subtype of the target object; the first subtype is a first type of subdivision type.

In the embodiment of the application, in view of the limitation of type detection when an image detection technology is used for detecting a target object in practical application, that is, the type of the target object cannot be subdivided, in the embodiment of the application, when the target object is detected to be of a first type, the intelligent driving assistance system performs second attribute extraction on a region to be processed to obtain second attribute information of the target object, and the intelligent driving assistance system can determine a first subtype of the target object according to the second attribute information; the first type is a rough subdivision type of the target object, a plurality of subdivision types are arranged under the first type, and the first subtype is one of the plurality of subdivision types.

Therefore, when the subdivision type of the target object cannot be directly determined, the target object can be understood by acquiring the second attribute information of the target object, so that the intelligent driving assistance system can make a safer and more reliable decision, and the driving safety is improved.

In a specific application scenario, such as an automatic driving application scenario, when the target object is a traffic sign, the second attribute information may include dominant color information; correspondingly, the performing of the second attribute extraction on the region to be processed to obtain the second attribute information of the target object includes: and extracting the main color of the area to be processed to obtain the main color information of the target object.

It should be noted that the dominant color is only one possibility of the second attribute, and in practical applications, an appropriate attribute having an ability of distinguishing a type of the target object is selected as the second attribute in combination with an object actually characterized by the scene and the target object.

In some possible embodiments, the performing the dominant color extraction on the region to be processed to obtain the dominant color information of the target object may include the following steps as shown in fig. 5:

in step S501, a sub-region of interest in the region to be processed for each candidate color of the plurality of candidate colors is determined.

Generally, the image to be detected is an RGB image; however, it is difficult to distinguish each candidate color conveniently in the RGB color space, and in the HSV color space, the chromaticity H, the saturation S, and the lightness V can show the candidate colors in different value ranges, so that the sub-region of interest corresponding to each candidate color can be extracted more flexibly by using the numerical intuitiveness of the HSV color space.

Specifically, performing color space conversion on an RGB image corresponding to the area to be processed to obtain an HSV image corresponding to the area to be processed; determining the interesting subarea of each candidate color in the HSV image according to the value range of each candidate color in the HSV color space; the sub-region of interest corresponding to each candidate color may be a plurality of connected pixels, or may not be a plurality of fully connected pixels, that is, there may be a hole in the sub-region, and there is an isolated pixel outside.

In a specific application scenario, the target object is a traffic sign, and the traffic sign mainly relates to red, blue, yellow, orange, green, cyan, black and white colors; thus, the plurality of candidate colors may include any of the plurality of colors described above. Taking red as an example, the value range of red on an H channel in an HSV color space is [0,10], the value range on an S channel is [43,255], and the value range on a V channel is [46,255]; in practical application, the sub-region of interest may be screened based on the value range of any one of the channels, or the sub-region of interest may be screened by combining the value ranges of the three channels.

In step S503, edge extraction is performed on the sub-region of interest corresponding to each candidate color, so as to obtain edge information of each candidate color.

The edge refers to a portion of an image where local brightness and color change are significant, and is generally a junction between one region and another attribute region in the image. Thus, the edge information corresponding to each candidate color may also indicate that the candidate color actually characterizes the outline of the object. Optionally, before performing the edge extraction, a region connection process may be performed on the sub-region of interest to solve a situation that pixels originally belonging to the same sub-region (i.e., the same object) are divided.

Specifically, when extracting edge information corresponding to each candidate color, firstly determining a circumscribed rectangle of a sub-region of interest corresponding to the candidate color, performing binarization processing on an image part corresponding to the circumscribed rectangle to obtain a binarized image corresponding to the sub-region of interest of the candidate color, and then determining edge pixels of the candidate color based on the binarized image; thus, the above steps are repeated, and a plurality of pieces of edge information corresponding to a plurality of candidate colors one to one can be obtained.

In step S505, the coverage area of each candidate color is determined based on the edge information of each candidate color.

Specifically, after determining edge information, i.e., edge pixels, of each candidate color, the number of pixels surrounded by the edge pixels of the candidate color, i.e., the number of pixels inside the edge, is determined, and the number of pixels inside the edge is used as the coverage area of the candidate color.

In step S507, a dominant color is determined from the plurality of candidate colors based on the coverage area of each candidate color, and dominant color information of the target object is obtained.

Specifically, the number of pixels inside each candidate color edge is sorted from large to small or from small to large to obtain a sorting result; then, the candidate color with the largest number of pixels in the ranking result is determined as the dominant color.

Or, taking the candidate color with the number of pixels larger than or equal to the preset number of pixels in the number of pixels inside each candidate color edge as the main color; the preset pixel number is the number of pixels actually covered by the main color of the target object in the image at a fixed distance; the preset pixel number can be determined according to an empirical value, for example, when the vehicle is at a distance of 10 meters from the traffic sign, the number of pixels occupied by the red main color of the highest speed limit board in the image to be detected is 50, and then the preset pixel number can be 50.

In the above embodiment, when the traffic sign in the actual scene is subdivided, the traffic sign may be distinguished by the peripheral outline color, for example, when the peripheral outline color is red, the traffic sign is usually a prohibition-type sign, and when the peripheral outline color is blue, the traffic sign is usually an indication-type sign; based on the method, in the area to be processed, the edge information of each candidate color is detected, and then the inner part of the edge is taken as the coverage range of the candidate color based on the edge information, so that the main color determined based on the coverage range is the peripheral outline color of the traffic sign.

It should be noted that in other embodiments, the number of pixels covered by each sub-region of interest may also be determined according to the sub-region of interest corresponding to each candidate color obtained in step S501, and then the main color is determined based on the number of pixels covered by each sub-region of interest; thus, the determined main color is the color with the widest color distribution in the area to be processed; in practical application, the color distribution rule in the traffic sign board can be summarized, and the traffic sign board can be distinguished according to the color distribution rule.

Therefore, aiming at the situation that whether the traffic sign board is a forbidden sign or an indication sign cannot be accurately distinguished through a model in the related art, and only the two kinds of sign boards can be classified into other types, the first subtype of the traffic sign board is directly determined according to the main color of the traffic sign board after the main color of the traffic sign board is determined through the embodiment; for example, in the above embodiment, when the main color of the traffic sign is recognized as red, it is determined that the first subtype of the traffic sign is the prohibited-type sign; when the traffic sign is identified as having a dominant color of blue, a first subtype of the traffic sign is determined to be an indication-type sign.

In this way, the sub-region of interest of each candidate color is obtained by processing the region to be processed, and the dominant color information of the target object is further determined based on the edge information of each candidate color, and the dominant color information may be used to determine the subdivision type of the target object.

In other embodiments, in combination with the association between the second attribute and the first subtype in the actual application scenario, when the model cannot subdivide the target object and can only classify the target object into the first type, the second attribute information of the target object is extracted, and the first subtype of the target object is determined in combination with the second attribute information. Therefore, the method can make up the deficiency of the classification capability of the model, improve the comprehensiveness of the target object detection, and contribute to the control of the target object in the actual application scene.

In addition, considering that in practical applications, the background of the traffic sign may be a green tree, that is, the green tree wraps the traffic sign, when the maximum edge coverage area is taken as the color of the peripheral outline of the traffic sign by using the method of the above embodiment, for example, the color of the real peripheral outline in fig. 3 is red, but the main color finally recognized due to being wrapped by the green tree is green. Based on this, for the situation that the traffic sign background is a green tree, which may be incorrectly identified, the following rules are summarized by observing a large number of traffic signs identified as green:

if the traffic sign only takes the green tree as the background, the difference between the minimum circumscribed rectangle size of the outline of the green tree and the minimum circumscribed rectangle size of the outline of the sign is about 5 pixels;

if the traffic sign is green, the dimension of the minimum bounding rectangle of the label outline differs by more than 5 pixels from the dimension of the minimum bounding rectangle of the 2 nd largest outline in s.

Thus, in some possible embodiments, the candidate colors are ranked based on the coverage area of each candidate color, and a ranking result is obtained; when the first candidate color in the ranking result is the first preset color, specifically, the first preset color is green, the method of the present application further includes the following steps as shown in fig. 6:

in step S601, it is determined whether the candidate color ranked second in the ranking result is a second preset color. If the second candidate color is the second preset color, executing the steps S603-605; otherwise, steps S609 to 613 are executed.

Here, the second preset color may include red.

In step S603, the size of the first circumscribed rectangle corresponding to the first preset color and the size of the second circumscribed rectangle corresponding to the second preset color are determined, respectively.

Specifically, based on the edge pixels of the first preset color, a minimum circumscribed rectangle surrounding the edge pixels of the first preset color is determined, and a first circumscribed rectangle corresponding to the first preset color and the size of the first circumscribed rectangle are obtained; similarly, a second external rectangle corresponding to the second preset color and the size of the second external rectangle can be obtained.

In step S605, it is determined whether the size of the first circumscribed rectangle and the size of the second circumscribed rectangle satisfy a first preset condition; if the first preset condition is satisfied, executing step S607; otherwise, step S617 is performed.

In step S607, the dominant color is updated to the second preset color.

In step S609, it is determined whether the second ranked candidate color is a third preset color; if the second ranked candidate color is the third preset color, executing step S611; otherwise, step S617 is performed.

Here, the third preset color may include blue.

In step S611, the size of the first circumscribed rectangle corresponding to the first preset color and the size of the third circumscribed rectangle corresponding to the third preset color are determined respectively.

Specifically, based on the edge pixel of the third preset color, the minimum circumscribed rectangle surrounding the edge pixel is determined, and a third circumscribed rectangle corresponding to the third preset color and the size of the third circumscribed rectangle are obtained.

In step S613, it is determined whether the size of the first circumscribed rectangle and the size of the third circumscribed rectangle satisfy a second preset condition; if the second preset condition is satisfied, executing step S615; otherwise, step S617 is performed.

In step S615, the main color is updated to the third preset color.

In step S617, the main color is maintained as the first preset color.

Specifically, referring to the actual observation result, the first preset condition and the second preset condition may both include: the number of pixels of the second external rectangle corresponding to the second preset color in the length direction is less than that of the first external rectangle corresponding to the first preset color in the length direction by more than 5, and/or the number of pixels of the second external rectangle corresponding to the second preset color in the width direction is less than that of the first external rectangle corresponding to the first preset color in the width direction by more than 5.

In some possible embodiments, by means of the regularity of the existence of the traffic sign, the color attribute of the traffic sign can be determined directly according to the type after the type of the traffic sign is determined, and the color attribute is not used for determining the subdivision type of the traffic sign at the moment, so as to meet the requirement of subsequent application; for example, the background color of the traffic sign of the highest speed limit type is white, the background color of the traffic sign of the lowest speed limit type is blue, and the traffic sign of the speed limit releasing type is white.

Here, the highest speed limit type, the lowest speed limit type, and the speed limit removal type may be directly detected and identified by the image to be detected in step S201; namely, during model training, the model is trained into the traffic signboards capable of distinguishing different speed limit types through the labeled training images of the speed limit types. And for the subdivision types which cannot be distinguished by other models, marking the subdivision types as the first types during training, and extracting the second attributes.

In addition, when the traffic sign board is determined to be of the first type and the color of the peripheral outline of the traffic sign board is determined to be red, whether the traffic sign board belongs to a prohibition sign of an outer ring red plus oblique bar or belongs to a sign for prohibiting the parking of a vehicle can be further judged. Specifically, if the periphery is determined to be a ring shape, and meanwhile, the penultimate candidate color is determined to be blue based on the sorting result, and the blue area meets the requirement, the vehicle parking prohibition sign is determined; otherwise, it is a red circle prohibition flag.

When the traffic sign board is determined to be of the first type and the color of the peripheral outline is yellow, determining whether the shape of the peripheral outline is a hollow triangular ring shape or not; if the peripheral outline shape is a hollow triangular ring shape, the triangular warning board such as rain and snow is judged, and the ground color is determined to be black.

The embodiment of the application provides the identification logic for identifying the traffic sign board, so that the accurate type of the traffic sign board can be accurately positioned.

In some possible embodiments, the method of the embodiments of the present application may further include the steps of:

Here, when the target object is of the second type, it is indicated that third attribute information to be extracted is still present in the target object; the third attribute information is supplementary information of a second type.

In a specific application scenario, the second type may include at least one of the highest speed limit type, the lowest speed limit type, and the release speed limit type described above.

Considering the actual driving scene, the automatic driving needs to identify the type of the traffic sign board, and needs to know the speed limit, namely the speed limit value, after identifying the highest speed limit type, the lowest speed limit type and the speed limit releasing type; the speed limit value is used for intelligent driving control. Therefore, when the first attribute information indicates that the target object is any one of the highest speed limit type, the lowest speed limit type, and the release speed limit type, third attribute extraction is performed on the to-be-processed area to obtain a specific speed limit value.

Thus, the above-mentioned third attribute information may include character information; correspondingly, the extracting the third attribute of the region to be processed to obtain the third attribute information of the target object may include: and extracting characters from the area to be processed to obtain character information of the target object.

Specifically, the character extraction may use an existing character extraction algorithm model in the related field, such as a paddleocr model. Further, the character information recognized by the paddleocr model includes characters in various forms such as numbers, letters, texts, and the like; in the actual scene of the application, only a specific pixel value needs to be obtained, so that the character information of the target object identified by the paddleocr model can be screened to obtain information in a digital form; subsequently, only the digital information of the target object can be reserved, and redundant character information is eliminated, so that the storage space can be saved.

For example, after the type of the traffic sign in fig. 3 is determined to be the highest speed limit type, that is, the second type, character extraction is performed on the area to be processed, and a specific speed limit value of 60 can be obtained. Therefore, the intelligent driving assistance system can provide more effective information and is beneficial to making more reasonable and safer decisions.

Further, after the plurality of attribute information of the target object is obtained, the intelligent driving assistance system can subsequently determine the target point cloud of the target object in the point cloud data based on the position information of the target object in the image to be detected on the basis of the point cloud data and the registered image of the laser radar in combination with the point cloud data of the laser radar; then, a plurality of attribute information (types, main colors and characters) of the target object and the target point cloud are associated and stored, so that the intelligent driving assistance system can realize the fusion of a plurality of sensing systems, comprehensively know the information of the target object and be beneficial to the intelligent driving assistance system to make safer and more reasonable driving decisions.

In summary, in the embodiment of the present application, after the image is detected by the deep learning model to obtain the first attribute information, that is, the type, of the target object, the second attribute and the third attribute of the region where the target object is located are extracted by combining with the conventional algorithm to obtain the second attribute information and the third attribute information of the target object, so that the segmentation type of the target object can be further determined based on the second attribute information, the problem that the deep learning model cannot directly obtain the segmentation type of the target object from the image in the related art is solved, and in addition, more supplementary information about the type of the target object can be obtained based on the third attribute information. Therefore, the defect of the detection capability of the model can be made up, and the comprehensiveness and the accuracy of the target object information detection are improved.

Fig. 7 is a block diagram illustrating a target object attribute information determination apparatus according to an example embodiment. Referring to fig. 7, the apparatus includes a detection module 701, an acquisition module 702, a determination module 703, and an extraction module 704;

the detection module 701 is configured to perform target object detection on an image to be detected to obtain a target object detection result;

an obtaining module 702, configured to obtain, from a target object detection result, position information of a target object in an image to be detected and first attribute information of the target object when the target object detection result indicates that the target object exists; the first attribute information represents the type of the target object;

a determining module 703, configured to determine a to-be-processed area corresponding to the position information from the to-be-detected image;

the extracting module 704 is configured to, when the first attribute information includes the first type information, perform second attribute extraction on the region to be processed to obtain second attribute information of the target object; the second attribute information is used for determining first subtype information of the target object; the first sub-type corresponding to the first sub-type information is a subdivision type of the first type corresponding to the first type information.

In some possible embodiments, the extracting module 704 is further configured to, when the first attribute information indicates that the target object is of the second type, perform third attribute extraction on the region to be processed to obtain third attribute information of the target object; the third attribute information is supplementary information of a second type.

In some possible embodiments, the detecting module 701 is further configured to perform target object detection on the image to be detected according to the target object detection model, so as to obtain a target object detection result;

In some possible embodiments, the method further comprises a generation module of the target object detection model;

the generation module of the target object detection model is also used for acquiring a labeled first training image set; each frame of the first training image in the first training image set comprises a target object; constructing a preset machine learning model; performing target object detection training on a preset machine learning model according to the first training image set until an initial training end condition is met to obtain an initial target object detection model; automatically labeling each frame of second training image in the unlabeled second training image set according to the initial target object detection model to obtain a labeled second training image set; verifying the marked second training image set to obtain a verified second training image set; and training the initial target object model according to the verified second training image set until a preset training end condition is met, and obtaining a target object detection model.

performing second attribute extraction on the region to be processed to obtain second attribute information of the target object, wherein the second attribute extraction comprises the following steps:

In some possible embodiments, the extracting module 704 is further configured to determine a sub-region of interest in the region to be processed for each candidate color in the plurality of candidate colors; performing edge extraction on the interesting sub-region corresponding to each candidate color to obtain edge information of each candidate color; determining a coverage area of each candidate color based on the edge information of each candidate color; and determining a main color from the candidate colors based on the coverage area of each candidate color to obtain the main color information of the target object.

In some possible embodiments, the third attribute information includes character information; the extracting module 704 is further configured to perform character extraction on the region to be processed to obtain character information of the target object.

With regard to the apparatus in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be described in detail here.

FIG. 8 is a block diagram illustrating an electronic device for target object property information determination in accordance with an illustrative embodiment.

The electronic device may be a server or a terminal device, and its internal structure diagram may be as shown in fig. 8. The electronic device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic equipment comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program to run on the non-volatile storage medium. The network interface of the electronic device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement a target object attribute information determination method.

Those skilled in the art will appreciate that the structure shown in fig. 8 is a block diagram of only a portion of the structure relevant to the present disclosure, and does not constitute a limitation on the electronic device to which the present disclosure may be applied, and that a particular electronic device may include more or less components than those shown, or combine certain components, or have a different arrangement of components.

In an exemplary embodiment, there is also provided an electronic device including: a processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement a target object attribute information determination method as in the embodiments of the present application.

In an exemplary embodiment, a computer-readable storage medium is also provided, and when executed by a processor of an electronic device, enables the electronic device to execute a target object attribute information determination method in an embodiment of the present application.

In an exemplary embodiment, a computer program product is further provided, where the computer program product includes a computer program, the computer program is stored in a readable storage medium, and at least one processor of a computer device reads and executes the computer program from the readable storage medium, so that the computer device executes the target object attribute information determination method of the embodiment of the present application.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), rambus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

It will be understood that the present application is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims

1. A method for determining attribute information of a target object is characterized by comprising the following steps:

when the target object detection result indicates that a target object exists, acquiring position information of the target object in the image to be detected and first attribute information of the target object from the target object detection result; the first attribute information represents the type of the target object; the target object is an object that has an impact on intelligent driving decisions;

when the first attribute information indicates that the target object is of a first type, performing second attribute extraction on the region to be processed to obtain second attribute information of the target object; the second attribute information is used for determining a first subtype of the target object; the first subtype is a subdivision type of the first type.

2. The method of determining object attribute information of claim 1, further comprising:

when the first attribute information indicates that the target object is of a second type, performing third attribute extraction on the region to be processed to obtain third attribute information of the target object; the third attribute information is the supplementary information of the second type.

3. The method for determining the attribute information of the target object according to claim 1 or 2, wherein the step of performing target object detection on the image to be detected to obtain a target object detection result comprises:

performing target object detection on the image to be detected according to the target object detection model to obtain a target object detection result;

wherein when the target object detection result indicates the presence of the target object, the target object detection result includes coordinate information and size information of a target object detection frame; the coordinate information and the size information are used for determining the position information of the target object in the image to be detected.

4. The method according to claim 3, wherein the generation of the target object detection model includes:

acquiring a labeled first training image set; each frame of a first training image of the first set of training images includes the target object;

constructing a preset machine learning model;

performing target object detection training on the preset machine learning model according to the first training image set until an initial training end condition is met to obtain an initial target object detection model;

and training the initial target object model according to the verified second training image set until a preset training end condition is met, and obtaining the target object detection model.

5. The target object attribute information determination method according to claim 1, wherein the second attribute information includes dominant color information;

the second attribute extraction of the region to be processed to obtain second attribute information of the target object includes:

6. The method for determining the attribute information of the target object according to claim 5, wherein the extracting the dominant color of the region to be processed to obtain the dominant color information of the target object includes:

determining a sub-region of interest of each candidate color of a plurality of candidate colors in the region to be processed;

performing edge extraction on the interested sub-region corresponding to each candidate color to obtain edge information of each candidate color;

and determining a main color from the candidate colors based on the coverage area of each candidate color to obtain main color information of the target object.

7. The method according to claim 6, wherein determining a dominant color from the plurality of candidate colors based on the coverage area of each candidate color to obtain dominant color information of the target object comprises:

based on the coverage area of each candidate color, sequencing the candidate colors to obtain a sequencing result;

if the first candidate color in the ranking result is a first preset color, acquiring a second candidate color in the ranking result;

if the sorted second candidate color is a second preset color, respectively determining the size of a first external rectangle corresponding to the first preset color and the size of a second external rectangle corresponding to the second preset color;

and if the size of the first external rectangle and the size of the second external rectangle meet a first preset condition, taking the second preset color as a main color to obtain main color information of the target object.

8. The method according to claim 2, characterized in that the third attribute information includes character information;

the third attribute extraction of the region to be processed to obtain third attribute information of the target object includes:

9. A target object attribute information determination apparatus, characterized by comprising:

the extraction module is used for performing second attribute extraction on the area to be processed to obtain second attribute information of the target object when the first attribute information comprises first type information; the second attribute information is used for determining first subtype information of the target object; the first subtype corresponding to the first subtype information is a first type of subdivision type corresponding to the first type information.

10. An electronic device, comprising:

a processor;

a memory for storing the processor-executable instructions;

wherein the processor is configured to execute the instructions to implement the target object property information determination method of any one of claims 1-8.

11. A computer-readable storage medium, wherein instructions in the computer-readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the target object property information determining method of any one of claims 1-8.