Animal experiment image processing method and apparatus, and computer device and storage medium

Abstract

The present application relates to the technical field of image processing. Disclosed are an animal experiment image processing method and apparatus, and a computer device and a storage medium. The method comprises: acquiring an animal experiment image; on the basis of a trained first data model, a trained first image segmentation model and the animal experiment image, marking sensitive regions in the animal experiment image to obtain a first image; on the basis of a trained second data model and the first image, replacing an animal experiment object in the first image to obtain a second image; and on the basis of a trained second image segmentation model and the second image, performing harmless processing on sensitive regions in the second image to obtain a target image. Sensitive regions in an image can be identified accurately and processed in a targeted manner, such that ethical and privacy issues in scientific research and education are reduced to the greatest extent, thereby effectively balancing the contradiction between the scientific research and educational value of animal experimental images and the reduction of public discomfort.

Description

Animal experiment image processing method and device, computer device and storage medium TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to an animal experiment image processing method and device, a computer device and a storage medium. BACKGROUND

[0002] In recent years, with the rapid development of science and technology, animal experiments play an important role in medical research, drug development and biology. The animal image data generated by these animal experiments is an indispensable resource in research and teaching.

[0003] However, such images often involve sensitive content, which may cause public ethical concerns, privacy issues and discomfort, posing a challenge to scientific communication and public education. In addition, these images come from different sources and there is no unified processing standard, making it difficult to effectively compare them. Currently, the direct visual impact of images is reduced through blurring or simple filters, which to some extent reduces the discomfort of the public, but at the same time loses too many details of the original image, reducing the scientific value and educational effect of the image. And the existing technology lacks the ability to automatically detect bloody and inappropriate content in experimental videos, relying on manual annotation and processing, which is not only inefficient but also inaccurate. In addition, existing image processing and virtual image generation are mainly used for human images, lacking customized solutions for the unique needs of animal experiment images, and not considering the standardization needs of these virtual images in scientific research and education, nor can they effectively identify and process specific sensitive content in images.

[0004] SUMMARY

[0005] Therefore, it is necessary to address the technical problem of poor processing effect of sensitive content in animal experiment images in the prior art, and to provide an animal experiment image processing method, device, computer device and storage medium. In a first aspect, an animal experiment image processing method is provided, the method comprising:

[0006] obtaining an animal experiment image;

[0007] based on the trained first data model, the first image segmentation model and the animal experiment image, marking the sensitive area on the animal experiment image to obtain a first image;

[0008] based on the trained second data model and the first image, replacing the animal experiment object in the first image to obtain a second image;

[0009] based on the trained second image segmentation model and the second image, harmless processing the sensitive area of the second image to obtain a target image.

[0010] In a second aspect, an animal experiment image processing device is provided, and the device comprises:

[0011] An acquisition module is configured to acquire an animal experiment image.

[0012] A marking module is configured to mark a sensitive region on the animal experiment image based on a trained first data model, a first image segmentation model, and the animal experiment image, to obtain a first image.

[0013] A replacement module is configured to replace an animal experiment subject in the first image based on a trained second data model and the first image, to obtain a second image.

[0014] A processing module is configured to perform harmless processing on the sensitive region of the second image based on a trained second image segmentation model and the second image, to obtain a target image.

[0015] In a third aspect, a computer device is provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the animal experiment image processing method when executing the computer program.

[0016] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program implements the steps of the animal experiment image processing method when executed by a processor.

[0017] The animal experiment image processing method provided by the present application comprises the following steps: acquiring an animal experiment image, marking a sensitive region on the animal experiment image based on a trained first data model, a first image segmentation model, and the animal experiment image, to obtain a first image, replacing an animal experiment subject in the first image based on a trained second data model and the first image, to obtain a second image, and performing harmless processing on the sensitive region of the second image based on a trained second image segmentation model and the second image, to obtain a target image. The sensitive region in the image can be accurately identified and processed by the first data model, the first image segmentation model, the second data model, and the second image segmentation model, so as to minimize the ethical and privacy issues in scientific research and education. At the same time, the animal experiment subject in the first image is replaced, which not only meets the needs of scientific research and education, but also promotes a new way of scientific communication and public education, and effectively balances the contradiction between the scientific research and education value of the animal experiment image and the reduction of public discomfort. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are only some of the embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings.

[0019] Wherein:

[0020] Fig. 1 is an application environment diagram of an animal experiment image processing method in an embodiment;

[0021] Fig. 2 is a flowchart of an animal experiment image processing method in an embodiment;

[0022] Fig. 3 is a structural block diagram of an animal experiment image processing device in an embodiment;

[0023] Fig. 4 is a structural block diagram of a computer device in an embodiment;

[0024] Fig. 5 is a structural block diagram of a computer device in another embodiment. DETAILED DESCRIPTION

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application; the description and the drawings are to be regarded as illustrative in nature and are not intended to limit the application. The terminology used in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The description and the drawings are to be regarded as illustrative in nature and are not intended to limit the application, as defined by the appended claims.

[0026] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase "in an embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily all directed to the same embodiments, or alternative or alternative embodiments.

[0027] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0028] The animal experiment image processing method provided by the embodiment of the present application can be applied in the application environment as shown in FIG. 1, wherein the client 110 communicates with the server 120 through a network. The server 120 can receive an animal experiment image through the client 110. Then, the server 120 labels a sensitive area on the animal experiment image based on a trained first data model, a first image segmentation model and the animal experiment image, to obtain a first image. Then, the server 120 replaces an animal experiment object in the first image based on a trained second data model and the first image, to obtain a second image. Finally, the server 120 performs harmless processing on the sensitive area in the second image based on a trained second image segmentation model and the second image, to obtain a target image. The sensitive area in the image can be accurately identified and processed by the first data model, the first image segmentation model, the second data model and the second image segmentation model, so as to reduce the ethical and privacy problems in scientific research and education to the greatest extent. At the same time, the animal experiment object in the first image is replaced, which not only meets the needs of scientific research and education, but also promotes a new way of scientific communication and public education, and effectively balances the contradiction between the scientific research and education value of the animal experiment image and the reduction of public discomfort. The client 110 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The server 120 can be implemented by an independent server or a server cluster composed of multiple servers. The present application will be described in detail through specific embodiments.

[0029] Referring to FIG. 2, FIG. 2 is a flowchart of an animal experiment image processing method provided by an embodiment of the present application, which includes the following steps:

[0030] Step S101: An animal experiment image is acquired.

[0031] Step S102: A sensitive area on the animal experiment image is labeled based on a trained first data model, a first image segmentation model and the animal experiment image, to obtain a first image.

[0032] In this embodiment, the first data model and the first image segmentation model can both be composed of convolutional neural networks (CNNs). Upon receiving an animal experiment image, the first data model can automatically identify key features in the animal experiment image, such as the animal species, experimental equipment, and specific content that may cause public discomfort. To improve the accuracy of identification, data augmentation and transfer learning techniques can also be introduced during the training phase of the first data model. Data augmentation increases the generalization ability of the model by performing operations such as rotation, scaling, and cropping on the training images. Transfer learning uses a pre-trained network model on other large-scale image datasets as a starting point to accelerate and optimize the training process for animal experiment images. As an example, the first data model detects the location and area of blood and inappropriate content, and uses image segmentation techniques (such as U-Net, Mask R-CNN) that is the first image segmentation model to accurately identify and label the inappropriate content area, ultimately obtaining the first image with labeled image areas that need to be processed or replaced.

[0033] Step S103: Based on the trained second data model and the first image, the animal experiment object in the first image is replaced to obtain a second image.

[0034] The second data model can be an advanced artificial intelligence technology such as a generative adversarial network (GAN).

[0035] As an example, the second data model can generate a virtual animal image that is visually similar to the original animal experiment object but does not cause discomfort. The virtual animal database stores the characteristics of different species of animals, including fur texture, color distribution, etc., to ensure that the generated virtual image is visually similar to the real animal and simulates the real experimental scene visually, while avoiding the ethical and privacy issues that may arise from directly using real animal experiment images. The generated virtual animal image can replace the animal experiment object in the first image, resulting in a replaced first image as the second image.

[0036] Step S104: Based on the trained second image segmentation model and the second image, the sensitive area of the second image is processed harmlessly to obtain a target image.

[0037] In this embodiment, the second image segmentation model is used to accurately distinguish different objects and backgrounds in the second image, and to process the sensitive area specifically. The processing process will intelligently adjust the image properties according to the specific content of the second image, reduce the saturation of blood, and blur the animal living environment, thereby effectively reducing the discomfort caused by the image. Through blurring, color adjustment, and the application of specific filters, harmless processing of images other than virtual animal images in animal experiment images is achieved.

[0038] In an embodiment, the first data model is obtained by model training based on a convolutional neural network, the second data model is obtained by model training based on a generative adversarial network, the second image segmentation model is obtained by model training based on a U-Net network, and the first image segmentation model is obtained by model training based on a U-Net network.

[0039] In an embodiment, the step of marking the sensitive region on the animal experiment image based on the trained first data model, the first image segmentation model, and the animal experiment image comprises:

[0040] Step S201: identifying the sensitive region of the animal experiment image based on the first data model and the animal experiment image.

[0041] Specifically, the first data model is used to identify the sensitive region of the animal experiment image.

[0042] Step S202: marking the sensitive region on the animal experiment image based on the first image segmentation model.

[0043] Specifically, the first image segmentation model is used to mark the sensitive region on the animal experiment image.

[0044] As an example, the first step: mark the features and regions that need to be identified in the animal experiment images of different time periods in the sample video, including the animal itself, the animal's wounds and surgical treatment parts, experimental instruments and other sensitive parts, and these marks will be one-to-one corresponding to the images. The second step: the animal experiment images are processed through a series of refined preprocessing operations. Specifically, the animal experiment image cropping: automatically cropping the irrelevant parts of the edge of the animal experiment image, such as background noise and irrelevant information, and retaining the central or important experimental area, the cropping algorithm is based on image edge detection technology, such as Canny edge detection, to identify the outline of the main object and ensure that the key content is completely retained; animal experiment image rotation and alignment: automatically rotating and correcting the animal experiment image to ensure that the main body direction of the animal experiment object is consistent, and using image feature matching and geometric transformation algorithms, such as SIFT (Scale-Invariant Feature Transform) feature-based matching, to automatically detect and correct the tilt of the image; animal experiment image normalization: normalizing the animal experiment image, adjusting the image size to a uniform size, and standardizing the pixel intensity range to [0, 1] or [-1, 1]. The third step: set the feature extraction network, the present application uses a pre-trained convolutional neural network (CNNs), in this example, the ResNet-50 network is used, the network automatically learns the hierarchical feature representation of the image through multiple convolutional layers and pooling layers, from basic textures and shapes to complex objects and scenes, and the output of the last convolutional layer is converted into a fixed-length feature vector, which is a high-level representation of the image, containing rich information about the image content, for subsequent feature recognition and location positioning. The fourth step: input the preprocessed image with labels into the ResNet-50 network for training, set the sample size, update function, learning rate and other parameters for each training, through supervised learning, the model learns to accurately identify sensitive areas from complex backgrounds and give corresponding labels. The fifth step: use the ResNet-50 network to process each frame of animal experiment image to detect the position and area of blood and inappropriate content, and then use image segmentation technology (such as U-Net, Mask R-CNN) to accurately identify and label the inappropriate content area. By adjusting the sensitivity threshold of the model, the strictness and accuracy of sensitive content identification can be flexibly controlled to ensure that the processed images meet the ethical standards and retain as much scientific value as possible.

[0045] In an embodiment, the step of replacing the animal experiment object in the first image based on the trained second data model and the first image to obtain a second image includes:

[0046] Step 301: generating a virtual animal object of the animal experiment object on the animal experiment image based on the second data model;

[0047] Step 302: replacing the animal experimental object in the first image based on the virtual animal object, and taking the replaced first image as the second image.

[0048] In this embodiment, the second data model can employ GANs, aiming to create virtual images that are visually close to real animal experimental objects but do not contain sensitive content, and the generated style is as standardized as possible. First, a large number of animal experimental images are obtained, which are selected as the data set for training GANs. At this time, each output image is labeled with the area containing sensitive content that needs to be replaced by a virtual image. Second, set the generator state space parameters for designing virtual images. The generator will accept random noise as input, convert these noises through a series of convolutional layers, and gradually build the rough outline and features of the virtual image. By defining the state space as a normal distribution of finite elements, the virtual images generated by the generator are standardized and normalized. Third, define specific features to be added to the virtual image, such as hair texture, eyes, mouth, etc. Fourth, set the discriminator. The discriminator compares the differences between the virtual image and the real animal image, provides feedback to the generator, and indicates which aspects need to be improved according to the labeling requirements. Based on the feedback from the discriminator, the generator iteratively optimizes its internal parameters to improve the realism and detail of the virtual image. Fifth, after multiple rounds of adversarial training, when the discriminator has difficulty distinguishing the differences between the virtual image and the real image, the iteration process will end. Finally, quality assessment is performed on the animal virtual image to confirm that the virtual image has high realism and completely avoids sensitive content.

[0049] In an embodiment, the step of replacing the animal experimental object in the first image based on the virtual animal object, and taking the replaced first image as the second image further comprises:

[0050] Step S401: performing replacement quality assessment on the second image to obtain a quality assessment result, wherein the quality assessment result includes the similarity between the virtual animal object and the animal experimental object.

[0051] In an embodiment, the step of performing harmless processing on the sensitive area of the second image based on the trained second image segmentation model and the second image to obtain the target image comprises:

[0052] Step S501: performing image segmentation of different objects on the second image based on the second image segmentation model to obtain each sub-region image.

[0053] Step S502: performing type identification on the sub-region image, and based on the type of the sub-region image, processing the sub-region image using the harmless strategy corresponding to the type, wherein the harmless strategy includes Gaussian blur or motion blur, color adjustment, and specific filters.

[0054] In this embodiment, the purpose is to convert the background part of the animal experiment image into a harmless form. The first step is to use advanced deep learning models such as U-Net or Mask R-CNN as the second image segmentation model, and use the second image segmentation model to subdivide the complex animal experiment image background into multiple regions with no objects. Here, the object can refer to a complete image, such as complete blood, surgical incision, etc. in the animal experiment image. The second step is to perform harmless processing on the sensitive areas identified after the image is successfully segmented. The harmless processing uses the following methods: (1) blur processing: for regions directly showing sensitive content such as blood and surgical incision, use Gaussian blur or motion blur technology for processing. (2) color adjustment: adjust specific colors in the image, such as adjusting red elements such as blood to a softer shade, or changing the color completely to reduce visual impact. (3) apply specific filters: for specific sensitive areas, specific image filters such as watercolor, sketch, and other stylized filters can be applied to reduce the direct display of sensitive content. The third step is to evaluate the visual quality of the image through automated tools, including clarity, color authenticity, and the degree of masking of sensitive content, to ensure that the harmless processing achieves the desired effect. Finally, manual review is conducted to evaluate the preservation of image content from the perspective of scientific research and education, ensuring that the image after harmless processing still maintains sufficient scientific value and educational significance.

[0055] In an embodiment, the step of performing type identification on the sub-region image and processing the sub-region image based on the type of the sub-region image using the harmless strategy corresponding to the type includes:

[0056] Step S601: Quality evaluation is performed on the sub-region image after harmless strategy processing, and an evaluation result is obtained. The evaluation result includes image clarity, image color authenticity, and the degree of masking of sensitive content.

[0057] The animal experiment image processing method provided in the embodiment can accurately identify the sensitive region in the image and perform targeted processing through the first data model, the first image segmentation model, the second data model, and the second image segmentation model, so as to minimize the ethical and privacy problems in scientific research and education, and at the same time, the animal experiment object in the first image is replaced, which not only meets the needs of scientific research and education, but also promotes a new way of scientific communication and public education, and effectively balances the contradiction between the scientific research and education value of the animal experiment image and the reduction of public discomfort.

[0058] Referring to FIG. 3, in an embodiment, an animal experiment image processing apparatus is provided, and the apparatus includes: an acquisition module 10 configured to acquire an animal experiment image;

[0059] A marking module 20 is configured to mark a sensitive region on the animal experiment image based on a trained first data model, a first image segmentation model, and the animal experiment image, to obtain a first image.

[0060] A replacement module 30 is configured to replace an animal experiment object in the first image based on a trained second data model and the first image, to obtain a second image.

[0061] A processing module 40 is configured to perform harmless processing on the sensitive region of the second image based on a trained second image segmentation model and the second image, to obtain a target image.

[0062] In an embodiment, the first data model is obtained by model training based on a convolutional neural network, the second data model is obtained by model training based on a generative adversarial network, the second image segmentation model is obtained by model training based on a U-Net network, and the first image segmentation model is obtained by model training based on a U-Net network.

[0063] The marking module 20 is configured to identify the sensitive region of the animal experiment image based on the first data model and the animal experiment image.

[0064] The sensitive region is marked on the animal experiment image based on the first image segmentation model.

[0065] The replacement module 30 is configured to generate a virtual animal object of the animal experiment object on the animal experiment image based on the second data model.

[0066] The replacement module 30 is configured to replace the animal experiment object in the first image based on the virtual animal object, and take the replaced first image as the second image.

[0067] The replacement module 30 is configured to perform replacement quality evaluation on the second image to obtain a quality evaluation result, and the quality evaluation result includes a similarity between the virtual animal object and the animal experiment object.

[0068] The processing module 40 is configured to perform image segmentation of different objects on the second image based on the second image segmentation model to obtain each sub-region image.

[0069] The processing module 40 is configured to perform type identification on the sub-region image, and perform processing on the sub-region image by using a harmless strategy corresponding to the type of the sub-region image based on the type of the sub-region image, wherein the harmless strategy includes Gaussian blur or motion blur, color adjustment, and a specific filter.

[0070] The processing module 40 is configured to perform quality evaluation on the sub-region image processed by the harmless strategy to obtain an evaluation result, and the evaluation result includes image definition, image color authenticity, and a masking degree of sensitive content.

[0071] In an embodiment, a computer device is provided, which can be a server. An internal structure diagram of the computer device can be as shown in FIG. 4. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is configured to communicate with an external client through a network connection. The computer program is executed by the processor to implement functions or steps of a server side of an animal experiment image processing method.

[0072] In one embodiment, a computer device is provided, which can be a client, and an internal structure diagram thereof can be as shown in FIG. 5. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile 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 in the non-volatile storage medium to run. The network interface of the computer device is used to communicate with an external server through a network connection. The computer program is executed by the processor to implement the functions or steps of the client side of the animal experiment image processing method.

[0073] In one embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the following steps when executing the computer program:

[0074] obtaining an animal experiment image;

[0075] based on the trained first data model, the first image segmentation model and the animal experiment image, marking the sensitive area on the animal experiment image to obtain a first image;

[0076] based on the trained second data model and the first image, replacing the animal experiment object of the first image to obtain a second image;

[0077] based on the trained second image segmentation model and the second image, harmless processing the sensitive area of the second image to obtain a target image.

[0078] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the following steps:

[0079] obtaining an animal experiment image;

[0080] based on the trained first data model, the first image segmentation model and the animal experiment image, marking the sensitive area on the animal experiment image to obtain a first image;

[0081] based on the trained second data model and the first image, replacing the animal experiment object of the first image to obtain a second image;

[0082] based on the trained second image segmentation model and the second image, harmless processing the sensitive area of the second image to obtain a target image.

[0083] It should be noted that the above functions or steps that can be achieved by the computer readable storage medium or the computer device can correspond to the above-mentioned method embodiments, and the related descriptions of the server side and the client side are not described again to avoid repetition.

[0084] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, the processes of the above-mentioned embodiments can be included. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory. 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. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM) and the like.

[0085] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of functional units and modules is exemplified. In actual application, the above-mentioned functions can be completed by different functional units or modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the above-described functions.

[0086] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. An animal experiment image processing method characterized by comprising: The animal experiment image processing method comprises: An animal experiment image is acquired; Based on the trained first data model, the first image segmentation model and the animal experiment image, a sensitive area on the animal experiment image is marked to obtain a first image; Based on the trained second data model and the first image, an animal experiment object of the first image is replaced to obtain a second image; Based on the trained second image segmentation model and the second image, the sensitive area of the second image is harmlessly processed to obtain a target image.

2. The animal experiment image processing method according to claim 1, characterized by, The first data model is obtained based on model training of a convolutional neural network, the second data model is obtained based on model training of a generative adversarial network, the second image segmentation model is obtained based on model training of a U-Net network, and the first image segmentation model is obtained based on model training of a U-Net network.

3. The animal experiment image processing method according to claim 2, characterized by, The step of marking the sensitive area on the animal experiment image based on the trained first data model, the first image segmentation model and the animal experiment image to obtain the first image comprises: Based on the first data model and the animal experiment image, the sensitive area of the animal experiment image is identified; Based on the first image segmentation model, the sensitive area on the animal experiment image is marked.

4. The animal experiment image processing method according to claim 3, characterized by, The step of replacing the animal experiment object of the first image based on the trained second data model and the first image to obtain the second image comprises: Based on the second data model, a virtual animal object of the animal experiment object on the animal experiment image is generated; Based on the virtual animal object, the animal experiment object of the first image is replaced, and the replaced first image is taken as the second image.

5. The animal experiment image processing method according to claim 4, characterized by, The step of replacing the animal experiment object of the first image based on the virtual animal object, and taking the replaced first image as the second image further comprises: The second image is replaced for quality evaluation to obtain a quality evaluation result, and the quality evaluation result comprises a similarity between the virtual animal object and the animal experiment object.

6. The animal experiment image processing method according to claim 1, characterized by, The step of harmlessly processing the sensitive area of the second image based on the trained second image segmentation model and the second image to obtain the target image comprises: Based on the second image segmentation model, the second image is subjected to image segmentation of different objects to obtain each sub-region image; The sub-region image is subjected to type identification, and based on the type of the sub-region image, a harmless strategy corresponding to the type is adopted to process the sub-region image, wherein the harmless strategy comprises Gaussian blur or motion blur, color adjustment, and a specific filter.

7. The animal experiment image processing method according to claim 6, characterized by, The step of processing the sub-region image based on the type of the sub-region image and adopting the harmless strategy corresponding to the type further comprises: The sub-region image processed by the harmless strategy is subjected to quality evaluation to obtain an evaluation result, and the evaluation result comprises image definition, image color authenticity and masking degree of sensitive content.

8. An animal experiment image processing apparatus characterized by comprising: The animal experiment image processing device comprises an acquisition module configured to acquire an animal experiment image. a marking module, configured to mark a sensitive region on the animal experiment image based on the trained first data model, the first image segmentation model, and the animal experiment image, to obtain a first image; a replacing module, configured to replace an animal experiment object in the first image based on the trained second data model and the first image, to obtain a second image; a processing module, configured to perform harmless processing on the sensitive region in the second image based on the trained second image segmentation model and the second image, to obtain a target image.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the animal experiment image processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to implement the steps of the animal experiment image processing method according to any one of claims 1 to 7.

Images