Image processing method and device, electronic equipment and storage medium

Abstract

The invention provides an image processing method and device, electronic equipment and a storage medium. The image processing method comprises the following steps: acquiring one or more input images; performing first target detection on the one or more input images to detect one or more potential threat targets in a region of interest in the one or more input images; in response to the detected one or more potential threat targets, cutting the one or more input images to obtain one or more potential threat area images corresponding to the one or more potential threat targets; performing second target detection on the one or more potential threat area images to determine one or more sensitive information areas in the one or more potential threat area images; and performing fuzzification operation on one or more sensitive information areas in the one or more potential threat area images.

Description

Technical Field

[0001] The present disclosure relates to the field of image processing technologies, and particularly to an image processing method, an image processing apparatus, an electronic device, and a computer-readable storage medium. Background Art

[0002] The two-wheeler is equipped with a sentry mode to enhance the vehicle's safety, boost the owner's confidence, and promote the intelligent development of the two-wheeler. After the sentry mode of the two-wheeler is activated, the vehicle continuously monitors the vehicle and the surrounding environment through the camera configured on the vehicle itself. When abnormal behavior or potential threats are detected, the sentry mode automatically captures and records the images around the vehicle, and sends an alarm to the owner through the mobile phone APP or text message. The owner can control the vehicle to sound the horn, flash the lights, etc. through the mobile phone to timely respond to potential risks.

[0003] The methods described in this section are not necessarily methods that have been previously envisioned or adopted. Unless otherwise specified, no method described in this section should be considered prior art solely because it is included in this section. Similarly, unless otherwise specified, the problems mentioned in this section should not be considered to have been recognized in any prior art. Summary of the Invention

[0004] The present disclosure provides an image processing method, an electronic device, and a storage medium to desensitize sensitive information in an image and then upload it to the cloud.

[0005] According to an aspect of the present disclosure, there is provided an image processing method, including: obtaining one or more input images; performing first object detection on the one or more input images to detect one or more potential threat objects in one or more regions of interest in the one or more input images; in response to detecting the one or more potential threat objects, cropping the one or more input images to obtain one or more potential threat region images corresponding to the one or more potential threat objects; performing second object detection on the one or more potential threat region images to determine one or more sensitive information regions in the one or more potential threat region images; and performing a blurring operation on the one or more sensitive information regions in the one or more potential threat region images.

[0006] According to another aspect of the present disclosure, there is provided an image processing apparatus, including: an acquisition unit configured to acquire one or more input images; a potential threat area determination unit configured to perform a first object detection on the one or more input images to detect one or more potential threat objects in an area of interest in the one or more input images; an image cropping unit configured to crop the one or more input images in response to detecting the one or more potential threat objects to obtain one or more potential threat area images corresponding to the one or more potential threat objects; a sensitive information area determination unit configured to perform a second object detection on the one or more potential threat area images to determine one or more sensitive information areas in the one or more potential threat area images; and a desensitization unit configured to perform a blurring operation on the one or more sensitive information areas in the one or more potential threat area images.

[0007] According to another aspect of the present disclosure, there is provided an electronic circuit, including: a circuit configured to perform the steps of the above method.

[0008] According to another aspect of the present disclosure, there is provided an electronic device. The electronic device includes: a processor; and a memory storing a program, the program including instructions that, when executed by the processor, cause the processor to execute the above method.

[0009] According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing a program. The program includes instructions that, when executed by a processor of an electronic device, cause the electronic device to execute the above method.

[0010] According to another aspect of the present disclosure, there is provided a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the above method.

[0011] According to another aspect of the present disclosure, there is provided a two-wheeled vehicle. The two-wheeled vehicle includes the above electronic device.

[0012] According to an embodiment of the present disclosure, by performing secondary object detection based on pure vision to desensitize sensitive information, not only can the use of complex sensors be reduced, the cost be lowered, but also the generalization ability of the object detection model can be improved, the object detection accuracy can be enhanced, computing resources can be saved, and the device power consumption can be reduced, thereby achieving large-scale promotion on two-wheeled vehicles.

[0013] According to the embodiments described hereinafter, these and other aspects of the present disclosure will be apparent and will be elucidated with reference to the embodiments described hereinafter. Description of the Drawings

[0014] The accompanying drawings exemplarily illustrate embodiments and form part of the specification, and are used together with the written description of the specification to explain the exemplary embodiments of the embodiments. The illustrated embodiments are for illustrative purposes only and do not limit the scope of the claims. In all the drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0015] Figure 1 A flowchart showing an exemplary process of an image processing method according to an embodiment of the present disclosure;

[0016] Figure 2 An exemplary block diagram of an image processing apparatus according to an embodiment of the present disclosure is shown; and

[0017] Figure 3 It is a block diagram showing an example of an electronic device according to an exemplary embodiment of the present disclosure. Detailed implementation manners

[0018] In the present disclosure, unless otherwise specified, the terms "first", "second", etc. are used to describe various elements and are not intended to limit the positional relationship, timing relationship or importance relationship of these elements. Such terms are only used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of the element, and in certain cases, based on the context description, they may also refer to different instances.

[0019] In the description of various examples in the present disclosure, the terms used are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element may be one or more. In addition, the term "and / or" used in the present disclosure covers any one of the listed items and all possible combinations.

[0020] In the related art, the sentinel mode system requires some hardware devices such as complex sensors, and the detection method has a high cost, making it difficult to be widely promoted and applied on two-wheel vehicles; some detection methods lack the ability of intelligent processing and cannot accurately sense abnormal behaviors or potential threats, resulting in misjudgments of the sentinel system and false alarms; the existing detection methods may perform differently in different regions and different environments, with poor generalization ability and unable to adapt to diverse application scenarios; the existing perception algorithms usually have a high complexity and require a large amount of computing resources, which not only increases the energy consumption of the device but also may affect the normal operation of other functions.

[0021] To solve the above problems in the related art, the present disclosure provides a new image processing method. By performing secondary object detection based on pure vision to desensitize sensitive information, it can not only reduce the use of complex sensors and lower costs, but also improve the generalization ability of the object detection model, enhance the object detection accuracy, save computing resources, and reduce device power consumption, thereby achieving large-scale promotion on two-wheel vehicles. The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0022] Figure 1 The flowchart shows an exemplary process of the image processing method according to an embodiment of the present disclosure.

[0023] In step S102, one or more input images can be obtained.

[0024] In step S104, one or more input images can be subjected to first object detection to detect one or more potential threat objects in the region of interest in the one or more input images.

[0025] In step S106, in response to detecting one or more potential threat objects, the one or more input images can be cropped to obtain one or more potential threat region images corresponding to the one or more potential threat objects.

[0026] In step S108, the one or more potential threat region images can be subjected to second object detection to determine one or more sensitive information regions in the one or more potential threat region images.

[0027] In step S110, a blurring operation can be performed on the one or more sensitive information regions in the one or more potential threat region images.

[0028] Using the image processing method provided by the embodiments of the present disclosure, by performing secondary object detection based on pure vision to desensitize sensitive information, it can not only reduce the use of complex sensors and lower costs, but also reduce the amount of calculation and device power consumption by performing secondary object detection on the potential threat region images after image cropping, thereby achieving large-scale promotion on two-wheel vehicles.

[0029] The following details each step of method 100.

[0030] In step S102, one or more input images can be obtained.

[0031] In some embodiments, the one or more input images can be obtained through the camera of the two-wheel vehicle.

[0032] In some embodiments, one or more input images are acquired by a front-view camera, a middle-view camera, and a rear-view camera of the two-wheeler. In some examples, the front-view camera may be located at the front of the two-wheeler for capturing the front of the two-wheeler; the middle-view camera may be located on the dashboard of the two-wheeler for capturing the status of the rider or the seat; and the rear-view camera may be located at the rear of the two-wheeler for capturing the rear of the two-wheeler.

[0033] In step S104, a first object detection is performed on one or more input images to detect one or more potential threat objects in the region of interest in the one or more input images.

[0034] In some examples, in step S104, a first object detection may be performed on one or more input images, such as the images acquired from the front-view camera, the middle-view camera, and the rear-view camera. Among them, the first object detection is directed at one or more potential threat objects entering the region of interest in the above images. Under the sentry system, the region of interest may include the area within a certain distance around the two-wheeler, and people, vehicles, or other dangerous objects entering this area may be recognized as potential threat objects by the sentry system. The region of interest can be determined based on a predetermined position and a predetermined size in the image captured by the camera, such as the region with a predetermined size centered on the image. In the embodiments of the present disclosure, an object detection model trained based on a deep learning method may be used to perform the object detection task. Without departing from the principle of the embodiments of the present disclosure, other suitable methods may also be used to perform the object detection task. In some examples, further logical judgments may be made on the people, vehicles, or other objects detected in the region of interest based on predetermined rules, such as based on the duration of stay and the movement speed in the region of interest, etc., to determine whether to recognize them as potential threat objects.

[0035] In some embodiments, method 100 may further include using the detection result of the first object detection on one or more input images as a new sample for training the algorithm model of the first object detection to perform updated training on the algorithm model of the first object detection. Among them, the detection result includes a bounding box (bbox) and a label. In this way, the images marked with a bounding box and a label obtained during use can be used as the training data set for updated training.

[0036] Thus, relying on a large amount of newly acquired image data, such as images of potential threat objects such as pedestrians or vehicles at various angles collected in various weather conditions such as daytime, night, rainy days, and foggy days, to train and update the algorithm model of the first object detection, so that the first object detection algorithm model has better generalization ability and improves the object detection accuracy.

[0037] In step S106, in response to detecting one or more potential threat targets, one or more input images may be cropped to obtain one or more potential threat region images corresponding to the one or more potential threat targets.

[0038] In some examples, in response to detecting one or more potential threat targets in a region of interest within a certain range near the two-wheeler, one or more input images may be cropped by an image cropping algorithm, such as image edge cropping, image threshold cropping, deep learning-based image cropping, etc., to obtain one or more potential threat region images corresponding to the one or more potential threat targets. Among them, the potential threat region image may be the image within the target detection box of the potential threat target.

[0039] Thus, by cropping one or more input images to obtain smaller one or more potential threat region images, the computational load of subsequent secondary object detection can be reduced, thereby reducing the computational cost, improving the computational efficiency, reducing the device power consumption, so as to achieve popularization on two-wheelers.

[0040] In some embodiments, method 100 may further include, in response to detecting one or more potential threat targets, starting local video stream recording.

[0041] Thus, automatic acquisition of images is realized, and the acquired images can be used for target detection model training to further improve the generalization ability of the target detection model and enhance the target detection accuracy.

[0042] In step S108, one or more potential threat region images may be subjected to secondary object detection to determine one or more sensitive information regions in the one or more potential threat region images. Among them, according to the actual situation, the secondary object detection and the primary object detection may use the same or different model structures.

[0043] In some examples, the potential threat target may be a person or a vehicle, and among them, the human face and license plate belong to sensitive information and need to be subjected to secondary object detection.

[0044] In some examples, the potential threat targets detected from the images of the front-view camera and rear-view camera of the two-wheeler may be a person or a vehicle, and the sensitive information is the human face and license plate. The potential threat targets detected from the images of the middle-view camera of the two-wheeler may be a person, and the sensitive information is the human face.

[0045] In the technical solution of the present disclosure, the processing of collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0046] In some embodiments, method 100 may further include using the detection results of performing second object detection on one or more images of potential threat areas as new samples for training the algorithm model of the second object detection, so as to perform updated training on the algorithm model of the second object detection. The detection results include object monitoring frames and labels. In this way, the images marked with object detection frames and labels obtained during use can be used as the training data set for updated training.

[0047] Thus, relying on a large amount of newly collected image data, for example, collecting pictures of sensitive information such as human faces or license plates at various angles under various weather conditions such as daytime, night, rainy days, and foggy days, to train and update the algorithm model of the second object detection, so that the second object detection algorithm model has better generalization ability and improves the object detection accuracy.

[0048] In some embodiments, at least one of the algorithm models of the first object detection and the second object detection may be an object detection model based on a neural network, such as the YOLO series algorithms, the R-CNN series algorithms, etc.

[0049] In some embodiments, method 100 may further include determining the ReLU function as the activation function of the above neural network model.

[0050] In some embodiments, method 100 may further include performing graph optimization operations on the convolutional layer (Conv), batch normalization layer (BN), and activation function of the object detection model after establishing the object detection model.

[0051] In some embodiments, method 100 may further include performing model reparameterization on the object detection model to couple multiple operators corresponding to the convolutional layer, batch normalization layer, and activation function of the object detection model.

[0052] In some examples, the above object detection model is a YOLOv8 algorithm model.

[0053] Here, taking the object detection model as the YOLOv8 algorithm model as an example for illustration:

[0054] 1) Since the Sigmoid function used in the original YOLOv8 algorithm model includes exponential operations and is relatively complex, the formula of the Sigmoid function is as follows:

[0055]

[0056] Therefore, replace the Sigmoid function with a simpler ReLU function, and the formula of the ReLU function is as follows:

[0057]

[0058] As can be seen from formula (2), the ReLU function only selects data without performing complex exponential operations, which can reduce the computational complexity of the model and improve the inference performance of the model on the edge side.

[0059] 2) In the YOLOv8 algorithm model, the Conv+BN+ReLU structure accounts for more than 80%. This structure needs to access memory three times. Among them, the formula for the Conv layer is as follows:

[0060] y = w * x + b Formula (3)

[0061] Among them, y is the result of convolution calculation; w is the weight, that is, the parameters learned from a large amount of data; x is the input data, that is, the input image; b is the bias term.

[0062] The formula for the BN layer is as follows:

[0063]

[0064] Among them, the convolution calculation result y in the Conv layer is the input x of the BN layer, E(x) is the mean of x, is the standard deviation of x, and γ and β are adjustment coefficients.

[0065] The ReLU formula is as formula (2). By combining formulas (2)-(4), the Conv+BN+ReLU structure diagram can be optimized into a single operator. The formula for this single operator is as follows:

[0066]

[0067] Among them, μ B corresponds to the mean of x in the BN layer, corresponds to the standard deviation of x in the BN layer. Thus, optimizing the Conv+BN+ReLU structure diagram into a single operator only requires accessing memory once, which can improve the inference performance of the model.

[0068] 3) Quantifying the weight parameters and activation values of the model can significantly improve the performance of the neural network on the premise of partial precision loss. The quantization methods can include pre-training quantization (PTQ), quantization during training (QAT), etc. The specific quantization methods are not limited here.

[0069] In some examples, PTQ INT8 quantization can be adopted to map the weight parameters and activation values in the original YOLOv8 algorithm model from FP32 to INT8, and directly use INT8 calculation in the hardware, thereby improving the inference speed. The formula for PTQ INT8 quantization is as follows:

[0070]

[0071] Among them, r represents the original float32 value; Z represents the offset of the float32 value; S represents the scaling factor of float32; Round() represents the mathematical function of rounding to the nearest integer; q represents a quantized integer value.

[0072] Thus, by optimizing the object detection algorithm model, reducing the model size and the amount of computation, the inference speed at the edge side is improved, making it more suitable for popularization on two-wheeled vehicles.

[0073] In step S110, a blurring operation can be performed on one or more sensitive information regions in one or more potential threat region images.

[0074] In some examples, for the images to be uploaded to the cloud, it is necessary to blur the sensitive information regions (such as faces, license plates, etc.) therein, such as pixel interference, data encryption, masking processing, etc., to achieve desensitization.

[0075] In some embodiments, method 100 may further include, after performing a blurring operation on one or more sensitive information regions in one or more potential threat region images, uploading the one or more blurred input images to the cloud to trigger an alarm.

[0076] In some examples, the desensitized input images can be uploaded to the cloud and an alarm can be triggered to remind the vehicle owner that there are potential threat targets nearby.

[0077] According to some embodiments of the present disclosure, the image processing method may further include: preprocessing one or more images to convert the one or more images into one or more images in a preset format.

[0078] One or more images obtained in step S102 may not be directly applicable as inputs to the object detection model. Therefore, these images can be converted into a preset format (for example, converted into the yuv format through a Video Input module, or the images can be converted into a format suitable for the neural network model), so that they can be applied to the object detection model to achieve fast and accurate detection.

[0079] According to an embodiment of the present disclosure, there is also provided an image processing apparatus. Figure 2An exemplary block diagram of an image processing apparatus according to an embodiment of the present disclosure is shown. The image processing apparatus 200 may include an acquisition unit 210 configured to acquire one or more input images; a potential threat area determination unit 220 configured to perform a first object detection on the one or more input images to detect one or more potential threat objects in the regions of interest in the one or more input images; an image cropping unit 230 configured to crop the one or more input images in response to detecting the one or more potential threat objects to obtain one or more potential threat area images corresponding to the one or more potential threat objects; a sensitive information area determination unit 240 configured to perform a second object detection on the one or more potential threat area images to determine one or more sensitive information areas in the one or more potential threat area images; and a desensitization unit 250 configured to perform a blurring operation on the one or more sensitive information areas in the one or more potential threat area images.

[0080] Here, the operations of the above respective units of the image processing apparatus are respectively similar to the operations of steps S102 to S110 described above, and will not be elaborated herein.

[0081] According to another aspect of the present disclosure, there is also provided an electronic circuit including a circuit configured to perform the steps of the above method.

[0082] According to another aspect of the present disclosure, there is also provided an electronic device including: a processor; and a memory storing a program, the program including instructions that, when executed by the processor, cause the processor to execute the above method.

[0083] According to another aspect of the present disclosure, there is also provided a non-transitory computer-readable storage medium storing a program, the program including instructions that, when executed by a processor of an electronic device, cause the electronic device to execute the above method.

[0084] According to another aspect of the present disclosure, there is also provided a computer program product including a computer program that, when executed by a processor, implements the above method.

[0085] According to another aspect of the present disclosure, there is also provided a two-wheeled vehicle including the above electronic device.

[0086] See Figure 3, an electronic device 300 will now be described, which is an example of a hardware device (electronic device) to which various aspects of the present disclosure can be applied. The electronic device 300 can be any machine configured to perform processing and / or computing, and can be, but is not limited to, a workstation, a server, a desktop computer, a laptop computer, a tablet computer, a personal digital assistant, a robot, a smart phone, an in-vehicle computer, or any combination thereof. The above-described image processing method 100 can be implemented in whole or at least in part by the electronic device 300 or a similar device or system.

[0087] The electronic device 300 can include (possibly via one or more interfaces) elements connected to or communicating with a bus 302. For example, the electronic device 300 can include a bus 302, one or more processors 304, one or more input devices 306, and one or more output devices 308. The one or more processors 304 can be any type of processor, and can include, but are not limited to, one or more general-purpose processors and / or one or more dedicated processors (such as special processing chips). The input device 306 can be any type of device capable of inputting information to the electronic device 300, and can include, but are not limited to, a mouse, a keyboard, a touch screen, a microphone, and / or a remote control. The output device 308 can be any type of device capable of presenting information, and can include, but are not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The electronic device 300 can also include a non-transitory storage device 310, and the non-transitory storage device can be any storage device that is non-transitory and can implement data storage, including but not limited to a disk drive, an optical storage device, a solid-state memory, a floppy disk, a flexible disk, a hard disk, a magnetic tape, or any other magnetic medium, an optical disc or any other optical medium, a ROM (read-only memory), a RAM (random access memory), a cache memory, and / or any other memory chip or cartridge, and / or any other medium from which a computer can read data, instructions, and / or code. The non-transitory storage device 310 can be removable from the interface. The non-transitory storage device 310 can have data / programs (including instructions) / codes for implementing the above-described methods and steps. The electronic device 300 can also include a communication device 312. The communication device 312 can be any type of device or system that enables communication with external devices and / or with a network, and can include, but are not limited to, a modem, a network card, an infrared communication device, a wireless communication device, and / or a chipset, such as a BluetoothTM device, an 802.11 device, a Wi-Fi device, a Wi-Max device, a cellular communication device, and / or the like.

[0088] The electronic device 300 may further include a working memory 314, which may be any type of working memory that can store programs (including instructions) and / or data useful for the operation of the processor 304, and may include, but is not limited to, random access memory and / or read-only memory devices.

[0089] Software elements (programs) may be located in the working memory 314, including but not limited to an operating system 316, one or more application programs 318, drivers, and / or other data and code. Instructions for performing the above methods and steps may be included in one or more application programs 318, and the above image processing method 100 may be implemented by the processor 304 reading and executing the instructions of one or more application programs 318. More specifically, in the above image processing method 100, steps S102 - S110 may be implemented, for example, by the processor 304 executing an application program 318 having instructions for steps S102 - S110. In addition, other steps in the above image processing method 100 may be implemented, for example, by the processor 304 executing an application program 318 having instructions for performing the corresponding steps. The executable code or source code of the instructions of the software elements (programs) may be stored in a non-transitory computer-readable storage medium (such as the above storage device 310), and when executed, may be loaded into the working memory 314 (possibly compiled and / or installed). The executable code or source code of the instructions of the software elements (programs) may also be downloaded from a remote location.

[0090] It should also be understood that various variations may be made according to specific requirements. For example, custom hardware may also be used, and / or specific elements may be implemented using hardware, software, firmware, middleware, microcode, a hardware description language, or any combination thereof. For example, some or all of the disclosed methods and devices may be implemented by programming hardware (such as programmable logic circuits including field programmable gate arrays (FPGAs) and / or programmable logic arrays (PLAs)) using logic and algorithms according to the present disclosure, in assembly language or a hardware programming language such as VERILOG, VHDL, C++.

[0091] It should also be understood that the foregoing method may be implemented in a server-client mode. For example, the client may receive data input by the user and send the data to the server. The client may also receive data input by the user, perform a part of the foregoing method processing, and send the processed data to the server. The server may receive data from the client and execute the foregoing method or another part of the foregoing method, and return the execution result to the client. The client may receive the execution result of the method from the server and, for example, present it to the user through an output device.

[0092] It should also be understood that the components of the electronic device 300 can be distributed over a network. For example, one processor can be used to perform some processing, while another processor located far from the one processor can perform other processing. Other components of the computing system 300 can be similarly distributed. In this way, the electronic device 300 can be interpreted as a distributed computing system that performs processing at multiple locations.

[0093] Some exemplary aspects of the present disclosure are described below.

[0094] Aspect 1. An image processing method, comprising:

[0095] Obtaining one or more input images;

[0096] Performing a first object detection on the one or more input images to detect one or more potential threat objects in the regions of interest in the one or more input images;

[0097] In response to detecting the one or more potential threat objects, cropping the one or more input images to obtain one or more potential threat region images corresponding to the one or more potential threat objects;

[0098] Performing a second object detection on the one or more potential threat region images to determine one or more sensitive information regions in the one or more potential threat region images; and

[0099] Performing a blurring operation on the one or more sensitive information regions in the one or more potential threat region images.

[0100] Aspect 2. The image processing method according to Aspect 1, further comprising:

[0101] Using the detection result of the first object detection on the one or more input images as a new sample for training the algorithm model of the first object detection, to update and train the algorithm model of the first object detection.

[0102] Aspect 3. The image processing method according to Aspect 1, further comprising:

[0103] Using the detection result of the second object detection on the one or more potential threat region images as a new sample for training the algorithm model of the second object detection, to update and train the algorithm model of the second object detection.

[0104] Aspect 4. The image processing method according to any one of Aspects 1 to 3, wherein the one or more input images are obtained through a camera of a two-wheeler.

[0105] Aspect 5. The image processing method according to Aspect 4, wherein the one or more input images are acquired by a front view camera, a middle view camera, or a rear view camera of a two-wheeler.

[0106] Aspect 6. The image processing method according to any one of Aspects 1 to 3, further comprising:

[0107] In response to detecting the one or more potential threat targets, starting local video stream recording.

[0108] Aspect 7. The image processing method according to any one of Aspects 1 to 3, wherein at least one of the algorithm models for the first object detection and the algorithm model for the second object detection is an object detection model based on a neural network.

[0109] Aspect 8. The image processing method according to Aspect 7, further comprising:

[0110] Determining the ReLU function as the activation function of the model of the neural network.

[0111] Aspect 9. The image processing method according to Aspect 8, further comprising:

[0112] After establishing the object detection model, performing graph optimization operations on the convolutional layer, batch normalization layer, and activation function of the object detection model.

[0113] Aspect 10. The image processing method according to Aspect 9, further comprising:

[0114] Performing model reparameterization on the object detection model to couple multiple operators corresponding to the convolutional layer, the batch normalization layer, and the activation function of the object detection model.

[0115] Aspect 11. The image processing method according to any one of Aspects 1 to 3, further comprising:

[0116] After performing a blurring operation on the one or more sensitive information regions in the one or more potential threat region images, uploading the blurred one or more input images to the cloud to trigger an alarm.

[0117] Aspect 12. An image processing apparatus, comprising:

[0118] An acquisition unit configured to acquire one or more input images;

[0119] A potential threat region determination unit configured to perform a first object detection on the one or more input images to detect one or more potential threat targets in one or more regions of interest in the one or more input images;

[0120] An image cropping unit, configured to crop the one or more input images in response to detecting the one or more potential threat targets, so as to obtain one or more potential threat area images corresponding to the one or more potential threat targets;

[0121] A sensitive information area determination unit, configured to perform a second target detection on the one or more potential threat area images to determine one or more sensitive information areas in the one or more potential threat area images; and

[0122] A desensitization unit, configured to perform a blurring operation on the one or more sensitive information areas in the one or more potential threat area images.

[0123] Aspect 13. An electronic circuit, comprising:

[0124] A circuit configured to execute the steps of the method according to any one of Aspects 1 to 11.

[0125] Aspect 14. An electronic device, comprising:

[0126] A processor; and

[0127] A memory storing a program, the program including instructions that, when executed by the processor, cause the processor to execute the method according to any one of Aspects 1 to 11.

[0128] Aspect 15. A non-transitory computer-readable storage medium storing a program, the program including instructions that, when executed by a processor of an electronic device, cause the electronic device to execute the method according to any one of Aspects 1 to 11.

[0129] Aspect 16. A computer program product, comprising a computer program, wherein the computer program, when executed by a processor, implements the method according to any one of Aspects 1 to 11.

[0130] Aspect 17. A two-wheeled vehicle, comprising the electronic device according to Aspect 14.

[0131] Although the embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it should be understood that the above methods, systems, and devices are merely exemplary embodiments or examples, and the scope of the present invention is not limited by these embodiments or examples, but is only defined by the authorized claims and their equivalent scope. Various elements in the embodiments or examples may be omitted or replaced by their equivalent elements. In addition, the steps may be executed in an order different from that described in the present disclosure. Further, various elements in the embodiments or examples may be combined in various ways. Importantly, with the evolution of technology, many of the elements described herein may be replaced by equivalent elements that emerge after the present disclosure.

Claims

1. An image processing method, comprising: Obtaining one or more input images; Performing first target detection on the one or more input images to detect one or more potential threat targets in the region of interest in the one or more input images; Responsive to detecting the one or more potential threat targets, cropping the one or more input images to obtain one or more potential threat region images corresponding to the one or more potential threat targets; Performing second target detection on the one or more potential threat region images to determine one or more sensitive information regions in the one or more potential threat region images; And Performing a blurring operation on the one or more sensitive information regions in the one or more potential threat region images.

2. The image processing method according to claim 1, further comprising: Using the detection result of performing the first target detection on the one or more input images as a new sample for training the algorithm model of the first target detection to update and train the algorithm model of the first target detection.

3. The image processing method according to claim 1, further comprising: Using the detection result of performing the second target detection on the one or more potential threat region images as a new sample for training the algorithm model of the second target detection to update and train the algorithm model of the second target detection.

4. The image processing method according to any one of claims 1 to 3, wherein, The one or more input images are obtained by a camera of a two-wheeler.

5. An image processing apparatus, comprising: An obtaining unit configured to obtain one or more input images; A potential threat region determining unit configured to perform first target detection on the one or more input images to detect one or more potential threat targets in the region of interest in the one or more input images; An image cropping unit configured to, responsive to detecting the one or more potential threat targets, crop the one or more input images to obtain one or more potential threat region images corresponding to the one or more potential threat targets; A sensitive information region determining unit configured to perform second target detection on the one or more potential threat region images to determine one or more sensitive information regions in the one or more potential threat region images; And A desensitization unit configured to perform a blurring operation on the one or more sensitive information regions in the one or more potential threat region images.

6. An electronic circuit, comprising: A circuit configured to execute the steps of the method according to any one of claims 1 to 4.

7. An electronic device, comprising: A processor; And A memory storing a program, the program including instructions that, when executed by the processor, cause the processor to execute the method according to any one of claims 1 to 4.

8. A non-transitory computer-readable storage medium storing a program, the program including instructions that, when executed by a processor of an electronic device, cause the electronic device to execute the method according to any one of claims 1 to 4.

9. A computer program product, comprising a computer program, wherein, The computer program, when executed by a processor, implements the method according to any one of claims 1 to 4.

10. A two-wheeled vehicle comprising the electronic device according to claim 7.

Images