Image data automatic labeling method and system based on deep learning

Abstract

The invention belongs to the technical field of computer vision, and discloses an automatic image data labeling method and system based on deep learning. The method comprises the following steps: constructing an image data automatic labeling model by using a deep learning algorithm, and constructing an automatic labeling strategy adjustment model by using a reinforcement learning algorithm; acquiring real-time performance data of the image data automatic labeling model, and inputting the real-time performance data into the automatic labeling strategy adjustment model; according to the real-time performance data, using an automatic annotation strategy adjustment model to adjust an automatic annotation strategy of the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model; and collecting real-time image data, and automatically labeling the real-time image data by using the adjusted image data automatic labeling model to obtain automatically labeled real-time image data. According to the method, the problems of low labeling efficiency, low labeling accuracy and low intelligent degree in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the field of computer vision technology, and specifically relates to a method and system for automatic labeling of image data based on deep learning. Background Art

[0002] Computer vision is an important branch of artificial intelligence, which aims to enable computers to "see" and "understand" the content in images and videos. With the development of computer vision technology, how to efficiently and automatically annotate image data has become an important development direction for computer vision applications. The demand for image data annotation has promoted the further innovation and application of computer vision technology. With the continuous advancement of technology, image annotation technology will become more efficient and accurate, and better meet the needs of different fields and scenarios.

[0003] The existing image data annotation technology has the following defects: 1) Low annotation efficiency: Existing image annotation methods often rely on manual work, which is not only time-consuming and labor-intensive, but also inefficient on large-scale image datasets and difficult to meet the needs of fast annotation; 2) Low annotation accuracy: Manual annotation is easily affected by subjective factors, such as the annotator’s experience and fatigue, which leads to errors and inconsistencies in the annotation results; 3) Low intelligence: Existing image annotation technologies lack the ability to dynamically adjust annotation strategies. Once the annotation performance degrades, they cannot adjust the annotation strategy in time to adapt to the new data distribution. Summary of the invention

[0004] In order to solve the problems of low labeling efficiency, low labeling accuracy and low intelligence in the prior art, the present invention aims to provide a method and system for automatic labeling of image data based on deep learning.

[0005] The technical solution adopted by the present invention is: A method for automatic annotation of image data based on deep learning, comprising the following steps: Use deep learning algorithms to build an automatic image data annotation model, and use reinforcement learning algorithms to build an automatic annotation strategy adjustment model; According to the preset model adjustment mechanism, real-time performance data of the image data automatic annotation model is collected, and the real-time performance data is input into the automatic annotation strategy adjustment model; According to the real-time performance data, the automatic annotation strategy of the automatic annotation strategy adjustment model is adjusted using the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model; Real-time image data is collected, and the adjusted image data automatic annotation model is used to automatically annotate the real-time image data to obtain automatically annotated real-time image data.

[0006] Furthermore, a deep learning algorithm is used to construct an automatic annotation model for image data, and a reinforcement learning algorithm is used to construct an automatic annotation strategy adjustment model, including the following steps: Collecting a number of historical image data, and preprocessing the number of historical image data to obtain a number of preprocessed historical image data; Based on some pre-processed historical image data, a deep learning algorithm is used to build an automatic image data annotation model; A model adjustment mechanism is set, and according to the model adjustment mechanism, image data is collected to automatically annotate a number of historical performance data of the model, and preprocessing is performed to obtain a number of preprocessed historical performance data; Based on some preprocessed historical performance data, a reinforcement learning algorithm is used to build an automatic labeling strategy adjustment model.

[0007] Furthermore, an automatic labeling model for image data is constructed based on an FPN-Faster R-CNN-DANN-DBN algorithm, and the automatic labeling model for image data includes a graph feature extraction module constructed based on the FPN algorithm, a target area positioning module constructed based on the Faster R-CNN algorithm, a domain adversarial training module constructed based on the DANN algorithm, and an automatic labeling module constructed based on the DBN algorithm. The graph feature extraction module, the target area positioning module, and the automatic labeling module are connected in sequence, the domain adversarial training module is connected to the target area positioning module, and the domain adversarial training module includes a label predictor and a domain classifier connected in sequence.

[0008] Furthermore, based on some pre-processed historical image data, a deep learning algorithm is used to construct an automatic image data annotation model, including the following steps: Use the CNN algorithm to build the basic network architecture of the graph feature extraction module; the basic network architecture includes alternating convolutional layers and pooling layers; Use the FPN algorithm to build a feature pyramid, and connect the outputs of all convolutional layers to the feature pyramid horizontally in a top-down order to obtain the initial graph feature extraction module; Use the Faster R-CNN algorithm to build the initial target area positioning module, and use the DANN algorithm to build the initial domain adversarial training module; Use the DBN algorithm to build the initial automatic annotation module; Integrate the initial graph feature extraction module, the initial target area positioning module, the initial domain adversarial training module and the initial automatic annotation module to obtain the initial image data automatic annotation model; Combining the loss functions of the initial graph feature extraction module, the initial target region localization module, the initial domain adversarial training module, and the initial automatic labeling module, a comprehensive loss function is obtained; Optimizing initial model parameters of the initial image data automatic annotation model to obtain an optimized image data automatic annotation model; Based on the comprehensive loss function, several preprocessed historical image data are used to optimize the training of the optimized image data automatic annotation model to obtain the final image data automatic annotation model.

[0009] Furthermore, taking minimizing the model error as the optimization goal, a swarm intelligence optimization algorithm is used to optimize the initial model parameters of the initial image data automatic annotation model to obtain an optimized image data automatic annotation model. Furthermore, the automatic labeling strategy adjustment model is constructed based on the MPO-PPO algorithm, and the automatic labeling strategy adjustment model includes a meta-strategy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the PPO algorithm. The reinforcement learning module is provided with an intelligent agent, a strategy network and an experience replay pool. The intelligent agent is respectively connected to the strategy network, the experience replay pool and the meta-strategy optimization module, and the meta-strategy optimization module is connected to the experience replay pool.

[0010] Furthermore, based on some pre-processed historical performance data, a reinforcement learning algorithm is used to build an automatic labeling strategy adjustment model, including the following steps: Use the FCM clustering algorithm to cluster a number of pre-processed historical performance data to obtain a number of cluster centers and corresponding cluster clusters, and set a historical strategy adjustment category for each cluster cluster; Use the experience replay mechanism to initialize the experience replay pool, build the agent of the reinforcement learning module, and define the state space, action space and reward function of the agent; The historical automatic labeling strategy adjustment problem is used as a simulation environment, the PPO algorithm is used to build a strategy network, and the initial reinforcement learning module is obtained by combining the experience replay pool and the intelligent agent. Set several policy adjustment categories as several sub-scenarios of meta-policy optimization, and use the MPO algorithm to build an initial meta-policy optimization module; According to some pre-processed historical performance data, the initial reinforcement learning module is trained and optimized, and the historical policy network parameters of the policy network of the reinforcement learning module in the training optimization are extracted to obtain the final reinforcement learning module, and generate some historical reinforcement learning experiences; Based on several sub-scenarios and several historical policy network parameters, the initial meta-policy optimization module is trained and optimized to obtain the final meta-policy optimization module, and several historical meta-policy optimization experiences are generated; Several historical meta-strategy optimization experiences and several historical reinforcement learning experiences are stored in the experience replay pool of the final reinforcement learning module, and the final meta-strategy optimization module and the final reinforcement learning module are integrated to obtain an automatic labeling strategy adjustment model.

[0011] Further, according to the real-time performance data, the automatic annotation strategy of the automatic annotation strategy adjustment model is adjusted using the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model, including the following steps: Obtaining similarities between the real-time performance data and a number of cluster centers, and using the historical strategy adjustment category of the cluster center with the highest similarity as the real-time strategy adjustment category of the real-time performance data; According to the real-time strategy adjustment category, the corresponding real-time meta-strategy optimization experience is extracted from a number of historical meta-strategy optimization experiences in the experience playback pool of the automatic labeling strategy adjustment model; Based on the experience of real-time meta-strategy optimization, the meta-strategy optimization module is used to initialize the real-time policy network parameters of the policy network of the reinforcement learning module to obtain an updated policy network. Randomly extract a number of real-time reinforcement learning experiences from a number of historical reinforcement learning experiences in the experience replay pool, and update the action space of the reinforcement learning module according to the number of real-time reinforcement learning experiences to obtain an updated action space; According to the real-time performance data, the state space of the reinforcement learning module is updated to obtain an updated state space; Use the agent to control the updated policy network to generate the probability distribution of all possible actions in the action space corresponding to each state in the state space; The possible action with the highest probability distribution in the action space is taken as the execution action of the state, and the execution actions of all states in the state space are integrated to obtain several automatic labeling strategy adjustment actions; According to a number of automatic labeling strategy adjustment actions, the automatic labeling strategy of the automatic labeling strategy adjustment model is adjusted to obtain an adjusted image data automatic labeling model.

[0012] Furthermore, real-time image data is collected, and the adjusted image data automatic annotation model is used to automatically annotate the real-time image data to obtain the automatically annotated real-time image data, including the following steps: Collecting real-time image data, and preprocessing the real-time image data to obtain preprocessed real-time image data; Using the graph feature extraction module of the adjusted image data automatic annotation model, extract several real-time feature graphs of the preprocessed real-time image data, and perform feature graph fusion to obtain real-time fusion features; Using the adjusted image data to automatically annotate the model’s target region positioning module, perform the target region positioning module based on the real-time fusion features to obtain several real-time target regions; The automatic annotation module of the adjusted image data automatic annotation model is used to automatically annotate several real-time target areas according to the real-time fusion features to obtain the automatically annotated real-time image data.

[0013] A deep learning-based automatic image data annotation system is used to implement an automatic image data annotation method. The system includes a model building unit, a performance data acquisition unit, a model adjustment unit and an automatic annotation unit that are connected in sequence.

[0014] The beneficial effects of the present invention are: The present invention provides a method and system for automatic annotation of image data based on deep learning, which can automatically process image data and significantly reduce manual participation, thereby greatly improving the annotation efficiency, especially when processing large-scale image data sets; the automatic annotation model of image data constructed using the deep learning algorithm can more accurately identify and annotate targets in images, reduce human errors and subjective inconsistencies, improve the accuracy of annotation, and increase the processing speed of image data, which is suitable for application scenarios that require rapid response; the automatic annotation strategy adjustment model constructed using the reinforcement learning algorithm has improved the degree of intelligence, and dynamically adjusts the annotation strategy according to real-time performance data, making model updates and iterations simpler and more efficient. By combining reinforcement learning and meta-strategy optimization, the automatic annotation model can be better generalized to new data sets, improving the performance of the automatic annotation model in different scenarios, and the domain adversarial training module realizes the migration of cross-domain knowledge, improving the performance of the automatic annotation model in cross-domain image annotation tasks.

[0015] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flowchart of the method for automatic annotation of image data based on deep learning in the present invention.

[0017] Figure 2 It is a structural block diagram of the image data automatic annotation system based on deep learning in the present invention. DETAILED DESCRIPTION

[0018] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0019] Embodiment 1: like Figure 1 As shown, this embodiment provides an automatic labeling method for image data based on deep learning, including the following steps: S1: Use deep learning algorithms to build an automatic image data annotation model, and use reinforcement learning algorithms to build an automatic annotation strategy adjustment model, including the following steps: S1-1: collecting a number of historical image data, and preprocessing the number of historical image data to obtain a number of preprocessed historical image data; Preprocessing includes data cleaning, Gaussian denoising, image enhancement, and normalization of historical image data, which improves the quality of the data and provides data support for subsequent model construction; S1-2: Based on some pre-processed historical image data, use deep learning algorithms to build an automatic image data annotation model; The image data automatic annotation model is constructed based on the Feature Pyramid Networks (FPN)-Faster R-CNN-Domain-Adversarial Neural Network (DANN)-Deep Belief Network (DBN) algorithm, and the image data automatic annotation model includes a graph feature extraction module constructed based on the FPN algorithm, a target region positioning module constructed based on the Faster R-CNN algorithm, a domain adversarial training module constructed based on the DANN algorithm, and an automatic annotation module constructed based on the DBN algorithm. The graph feature extraction module, the target region positioning module, and the automatic annotation module are connected in sequence, the domain adversarial training module is connected to the target region positioning module, and the domain adversarial training module includes a label predictor and a domain classifier that are connected in sequence. The graph feature extraction module includes a basic network architecture based on the Convolutional Neural Networks (CNN) algorithm, extracts feature maps on feature maps of different scales, and fuses the feature maps of different levels extracted by CNN through the feature pyramid network structure using jump connections. Through upsampling and lateral connections, a feature pyramid is constructed to transfer high-level semantic information to low-level layers, enhancing the semantic expression ability of low-level features, which can effectively combine low-level detail features and high-level semantic features; the target area positioning module adopts Faster The R-CNN target detection algorithm includes a region proposal network and a classifier for locating the target region in the image; the domain adversarial training module is used to distinguish the features of the source domain and the target domain, and try to distinguish which domain these features come from. Through this adversarial training, the target region positioning module gradually learns to ignore the information related to the domain and only retains the information related to the task, thereby improving the prediction efficiency; the gradient reversal layer is located in the label predictor, and its function is to reverse the gradient during the back propagation process, so that the domain classifier can learn to distinguish the features of the source domain and the target domain during the training process, and the feature extraction part of the target region positioning module is trained to extract and learn domain-invariant features, thereby realizing adversarial training. The domain classifier is used to predict whether the feature belongs to the source domain or the target domain based on the inverted features output by the gradient reversal layer; the automatic labeling module generates label labels based on the fused features; Based on some pre-processed historical image data, a deep learning algorithm is used to build an automatic image data annotation model, including the following steps: S1-2-1: Use the CNN algorithm to build the basic network architecture of the graph feature extraction module; the basic network architecture includes alternately connected convolutional layers and pooling layers; through the alternating connection of convolutional layers and pooling layers, multi-scale feature maps can be effectively extracted from the image, which can capture the local features of the image, and the spatial dimension of the features can be reduced through the pooling layer, while maintaining important feature information; S1-2-2: Use the FPN algorithm to build a feature pyramid. Connect the outputs of all convolutional layers to the feature pyramid in a top-down order to obtain the initial graph feature extraction module. The feature pyramid helps improve the accuracy of object detection, especially in object detection at different scales. S1-2-3: Use the Faster R-CNN algorithm to build the initial target region positioning module, and use the DANN algorithm to build the initial domain adversarial training module; the Faster R-CNN algorithm improves the efficiency and accuracy of target detection and realizes target region positioning; the DANN algorithm enables the model to share feature representations between the source domain and the target domain through adversarial training. The label predictor and domain classifier are used to predict the target category and domain label respectively, which helps the model to transfer learning between different data distributions and improves the model's generalization ability on unseen data; S1-2-4: Use the DBN algorithm to build the initial automatic labeling module; pre-train the network in an unsupervised manner and then perform supervised fine-tuning to help improve the accuracy of labeling; S1-2-5: Integrate the initial graph feature extraction module, the initial target area positioning module, the initial domain adversarial training module and the initial automatic annotation module to obtain the initial image data automatic annotation model; S1-2-6: Combine the loss functions of the initial graph feature extraction module, the initial target region positioning module, the initial domain adversarial training module, and the initial automatic labeling module to obtain a comprehensive loss function; the comprehensive loss function can balance the importance of different modules and optimize the performance of the entire model; S1-2-7: Taking minimizing the model error as the optimization goal, the Improved SparrowSearch Algorithm (ISSA) algorithm is used to optimize the initial model parameters of the initial image data automatic annotation model to obtain an optimized image data automatic annotation model, including the following steps: S1-2-7-1: Minimizing the model error is used as the optimization goal of the ISSA algorithm, and the initial model parameters of the initial image data automatic annotation model are encoded as individual vectors of the ISSA algorithm; S1-2-7-2: Set the algorithm parameters of the ISSA algorithm, and set the fitness function of the ISSA algorithm according to the optimization goal; The algorithm parameters include ISSA population parameters, maximum number of iterations, and search space limits. In this embodiment, the ISSA population parameters include a search space of dimension, determine the search space food and the sparrow position is ;in, is the search space food matrix, They are all elements of the food matrix in the search space. is the sparrow position matrix, are all elements of the sparrow position matrix, is the ISSA population size, Optimize the dimensions of the problem for the model; h is the number of ISSA individuals; the search space limit includes the search space upper limit And the search space lower limit is ; T is the transpose symbol; The formula of the fitness function is:

[0020] In the formula, ISSA individual The fitness value of is the model error function; is the ISSA individual variable; c is the ISSA individual indicator; S1-2-7-3: Based on the algorithm parameters and individual vectors, the Circle chaotic mapping sequence is used to initialize the population and obtain an initial ISSA population including several initial ISSA individuals; The formula is:

[0021] In the formula, is the initial ISSA individual of Circle chaos mapping; is the randomly generated initial ISSA individual; c is the ISSA individual indicator; S1-2-7-4: According to the fitness function, using the ISSA algorithm, the initial ISSA population is assigned identities and updated to obtain an updated ISSA population including a number of updated ISSA individuals, and the optimal individual is retained, including the following steps: S1-2-7-4-1: Use the fitness function to obtain the fitness value of each initial ISSA individual in the initial ISSA population; S1-2-7-4-2: Sort the initial ISSA individuals according to their fitness values ​​to obtain the initial discoverers, initial joiners and initial predators; S1-2-7-4-3: Update the initial ISSA population to obtain an updated ISSA population; the updated ISSA population includes updated discoverers, updated joiners and updated predators; The update formula of the discoverer is:

[0022] In the formula, Respectively t +1, t The iteration cISSA individuals who discovered it; is the maximum number of iterations; is a random number between 0 and 1; is a normally distributed random number; for A matrix whose elements are all 1; is the warning value; is the safety threshold; The update formula for the joiner is:

[0023] In the formula, Respectively t +1, t The iteration c ISSA individuals who join; The best position for the exposed person to occupy; is the current worst position; is a random number between 0 and 1; for A matrix whose elements are all 1 or -1; h is the number of ISSA individuals; A + To update the parameters; The update formula of the predator is:

[0024] In the formula, Respectively t +1, t The iteration c predator ISSA individuals; is the step size control parameter, and , is the convergence factor, A positive real number that controls the step size that is not 0; is the current best position; are the current, best and worst fitness of ISSA individuals respectively; is the minimum constant to prevent the denominator from being 0;

[0025] In the formula, is the convergence factor; tanh(.) is the hyperbolic tangent function; , are the maximum and minimum values ​​of the convergence factor, respectively; λ is the deceleration rate parameter, is the decreasing cycle parameter, λ =-2 π , = π ; S1-2-7-4-5: Introduce the Gaussian mutation mechanism, perform Gaussian mutation on the updated ISSA population, obtain a Gaussian mutation ISSA population including several Gaussian mutation ISSA individuals, and retain the optimal individual; The formula is:

[0026] In the formula, ISSA individuals with Gaussian variation; G (1,1) is the Gaussian variation parameter; For the updated ISSA entity; S1-2-7-6: Introduce a dynamic reverse mechanism to dynamically reverse the updated ISSA population, obtain a dynamically reversed ISSA population including several dynamically reversed ISSA individuals, and retain the best individual; The formula is:

[0027] In the formula, is a dynamically reversed ISSA individual; is the decreasing inertia coefficient; is the upper limit of the search space; is the lower limit of the search space; For the updated ISSA entity; S1-2-7-7: If the current number of iterations reaches the maximum number of iterations or the fitness value of the optimal individual is lower than the fitness value threshold, the optimal individual is output and the next step is entered; S1-2-7-8: Decode the individual vector of the optimal individual to obtain the optimal initial model parameters of the initial image data automatic annotation model; S1-2-7-9: Optimizing the initial image data automatic labeling model according to the optimal initial model parameters to obtain an optimized image data automatic labeling model; The ISSA algorithm optimizes the image data automatic annotation model, avoids the sensitivity of the image data automatic annotation model to the initial value, improves the training efficiency and accuracy of the image data automatic annotation model, and avoids the image data automatic annotation model from falling into premature maturity; S1-2-8: Based on the comprehensive loss function, use a number of pre-processed historical image data to optimize and train the optimized image data automatic annotation model to obtain the final image data automatic annotation model; S1-3: setting a model adjustment mechanism, and according to the model adjustment mechanism, collecting image data to automatically annotate a number of historical performance data of the model, and performing preprocessing to obtain a number of preprocessed historical performance data; S1-4: Based on some pre-processed historical performance data, use the reinforcement learning algorithm to build an automatic labeling strategy adjustment model; The automatic labeling strategy adjustment model is constructed based on the Meta-Policy Optimization (MPO)-Proximal Policy Optimization (PPO) algorithm, and the automatic labeling strategy adjustment model includes a meta-policy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the PPO algorithm. The reinforcement learning module is provided with an intelligent agent, a policy network, and an experience replay pool. The intelligent agent is connected to the policy network, the experience replay pool, and the meta-policy optimization module respectively, and the meta-policy optimization module is connected to the experience replay pool. Based on some pre-processed historical performance data, a reinforcement learning algorithm is used to build an automatic labeling strategy adjustment model, which includes the following steps: S1-4-1: Use the Fuzzy-C Means (FCM) clustering algorithm to cluster several pre-processed historical performance data to obtain several cluster centers and corresponding cluster clusters, and set a historical strategy adjustment category for each cluster cluster; clustering helps to identify different strategy adjustment categories in performance data and provide a basis for subsequent meta-strategy optimization; S1-4-2: Use the experience replay mechanism to initialize the experience replay pool, build the agent of the reinforcement learning module, and define the state space, action space and reward function of the agent; the experience replay pool can break the correlation between data and improve the stability and efficiency of learning; The experience replay pool is used to store meta-strategy optimization experience and reinforcement learning experience. Reinforcement learning experience includes state, action, reward, and next state. Meta-strategy optimization experience includes network parameters of the policy network in the reinforcement learning module and the corresponding policy adjustment category. S1-4-3: Use the historical automatic labeling strategy adjustment problem as a simulation environment, use the PPO algorithm, build a policy network, and combine the experience replay pool and the intelligent agent to obtain the initial reinforcement learning module; PPO is a reinforcement learning algorithm that optimizes the parameters of the policy network to maximize the cumulative reward, which can effectively balance exploration and utilization while maintaining the stability of training; S1-4-4: Set several policy adjustment categories as several sub-scenarios of meta-policy optimization, and use the MPO algorithm to build the initial meta-policy optimization module; allow the model to learn different policies in different scenarios, improve the adaptability and generalization ability of the policy, and meta-policy optimization can enable the model to quickly adapt to different tasks or environments without having to learn from scratch; S1-4-5: Based on some pre-processed historical performance data, the initial reinforcement learning module is trained and optimized, and the historical policy network parameters of the policy network of the reinforcement learning module in the training optimization are extracted to obtain the final reinforcement learning module, and generate some historical reinforcement learning experiences; S1-4-6: Based on several sub-scenarios and several historical policy network parameters, the initial meta-policy optimization module is trained and optimized to obtain the final meta-policy optimization module, and several historical meta-policy optimization experiences are generated; S1-4-7: storing several historical meta-strategy optimization experiences and several historical reinforcement learning experiences in the experience replay pool of the final reinforcement learning module, and integrating the final meta-strategy optimization module and the final reinforcement learning module to obtain an automatic labeling strategy adjustment model; S2: According to the preset model adjustment mechanism, real-time performance data of the image data automatic annotation model is collected, and the real-time performance data is input into the automatic annotation strategy adjustment model; The model adjustment mechanism includes the time period for collecting the performance data of the automatic annotation model of image data, and the performance indicators included in the performance data, such as accuracy, recall rate, F1 score, etc.; S3: According to the real-time performance data, the automatic annotation strategy of the automatic annotation strategy adjustment model is adjusted using the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model, including the following steps: S3-1: Obtain the similarity between the real-time performance data and several cluster centers, and use the historical strategy adjustment category of the cluster center with the highest similarity as the real-time strategy adjustment category of the real-time performance data; this helps to identify the strategy adjustment category to which the real-time data belongs, thereby selecting the most appropriate meta-optimization strategy; S3-2: According to the real-time strategy adjustment category, extract the corresponding real-time meta-strategy optimization experience from several historical meta-strategy optimization experiences in the experience playback pool of the automatic labeling strategy adjustment model; allowing the model to use past learning experience in similar situations to quickly adapt to new labeling scenarios; S3-3: Based on the experience of real-time meta-strategy optimization, the meta-strategy optimization module is used to initialize the real-time policy network parameters of the policy network of the reinforcement learning module to obtain an updated policy network; this ensures that the policy network can quickly adjust its behavior according to the current labeling task requirements; S3-4: Randomly extract a number of real-time reinforcement learning experiences from a number of historical reinforcement learning experiences in the experience replay pool, and update the action space of the reinforcement learning module based on the real-time reinforcement learning experiences to obtain an updated action space; the updated action space can reflect the possible actions of the current labeling task, and improve the adaptability and flexibility of the automatic labeling strategy; S3-5: Update the state space of the reinforcement learning module based on real-time performance data to obtain an updated state space; this helps the model better understand the current labeling task and make more appropriate decisions; S3-6: Use the agent to control the updated policy network to generate the probability distribution of all possible actions in the action space corresponding to each state in the state space; allowing the model to select the best action among multiple possible actions to maximize the expected reward; S3-7: Take the possible action with the highest probability distribution in the action space as the execution action of the state, integrate the execution actions of all states in the state space, and obtain several automatic labeling strategy adjustment actions; this helps the model make more accurate and effective decisions during the automatic labeling process; Automatic labeling strategy adjustment actions include: The weight redistribution action adjusts the weights of different parts in the comprehensive loss function of the image data automatic annotation model. For example, if the detected target area has many positioning errors, the weight of the positioning loss in the total loss can be increased; Feature selection optimization action: Reselect or optimize the features used for annotation based on the specific situation of performance degradation, which may include adding new features or discarding irrelevant features; Sample re-labeling action: re-label the image data that the model has incorrectly labeled, and use these samples as part of the training data to improve the model; S3-8: adjusting the automatic annotation strategy of the automatic annotation strategy adjustment model according to a number of automatic annotation strategy adjustment actions, and obtaining an adjusted image data automatic annotation model; S4: collecting real-time image data, and automatically annotating the real-time image data using the adjusted image data automatic annotation model to obtain the automatically annotated real-time image data, including the following steps: S4-1: collect real-time image data and preprocess the real-time image data to obtain preprocessed real-time image data; the preprocessing may include operations such as scaling, cropping, normalization, and denoising to ensure that the image data is suitable for input into the adjusted image data automatic annotation model, which can improve the quality of the image data, reduce the complexity of model processing, and help improve the accuracy and efficiency of subsequent annotation; S4-2: Using the graph feature extraction module of the automatic annotation model of the adjusted image data, extracting several real-time feature graphs of the preprocessed real-time image data, and performing feature graph fusion to obtain real-time fused features; through feature graph fusion, the model can capture detailed information and high-level structures in the image, which helps to locate and identify the target more accurately; S4-3: Use the adjusted image data to automatically annotate the target region positioning module of the model, and perform the target region positioning module according to the real-time fusion features to obtain several real-time target regions; accurate target region positioning is the key to automatic annotation, ensuring that the automatic annotation module can annotate the correct image region; S4-4: Using the automatic annotation module of the adjusted image data automatic annotation model, several real-time target areas are automatically annotated according to the real-time fusion features to obtain the real-time image data after automatic annotation; the output of the automatic annotation module is the image data with annotation information, which directly provides an explanation of the image content and is crucial for image analysis and understanding.

[0028] Embodiment 2: like Figure 2 As shown, this embodiment provides an automatic image data annotation system based on deep learning, which is used to implement an automatic image data annotation method. The system includes a model building unit, a performance data acquisition unit, a model adjustment unit and an automatic annotation unit connected in sequence; A model building unit, used to build an automatic annotation model for image data using a deep learning algorithm, and to build an automatic annotation strategy adjustment model using a reinforcement learning algorithm; A performance data collection unit, used to collect real-time performance data of the image data automatic annotation model according to a preset model adjustment mechanism, and input the real-time performance data into the automatic annotation strategy adjustment model; A model adjustment unit, used to adjust the automatic annotation strategy of the automatic annotation strategy adjustment model according to the real-time performance data, and obtain an adjusted image data automatic annotation model; The automatic annotation unit is used to collect real-time image data, and use the adjusted image data automatic annotation model to automatically annotate the real-time image data to obtain the automatically annotated real-time image data.

[0029] The present invention provides a method and system for automatic annotation of image data based on deep learning, which can automatically process image data and significantly reduce manual participation, thereby greatly improving the annotation efficiency, especially when processing large-scale image data sets; the automatic annotation model of image data constructed using the deep learning algorithm can more accurately identify and annotate targets in images, reduce human errors and subjective inconsistencies, improve the accuracy of annotation, and increase the processing speed of image data, which is suitable for application scenarios that require rapid response; the automatic annotation strategy adjustment model constructed using the reinforcement learning algorithm has improved the degree of intelligence, and dynamically adjusts the annotation strategy according to real-time performance data, making model updates and iterations simpler and more efficient. By combining reinforcement learning and meta-strategy optimization, the automatic annotation model can be better generalized to new data sets, improving the performance of the automatic annotation model in different scenarios, and the domain adversarial training module realizes the migration of cross-domain knowledge, improving the performance of the automatic annotation model in cross-domain image annotation tasks.

[0030] The present invention is not limited to the above optional implementations, and anyone can derive other various forms of products under the enlightenment of the present invention. The above specific implementations should not be understood as limiting the scope of protection of the present invention. The scope of protection of the present invention should be based on the definition in the claims, and the description can be used to interpret the claims.

Claims

1. A method for automatic annotation of image data based on deep learning, characterized in that: The steps include: Use deep learning algorithms to build an automatic image data annotation model, and use reinforcement learning algorithms to build an automatic annotation strategy adjustment model; According to the preset model adjustment mechanism, real-time performance data of the image data automatic annotation model is collected, and the real-time performance data is input into the automatic annotation strategy adjustment model; According to the real-time performance data, the automatic annotation strategy of the automatic annotation strategy adjustment model is adjusted using the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model; Real-time image data is collected, and the adjusted image data automatic annotation model is used to automatically annotate the real-time image data to obtain automatically annotated real-time image data.

2. The method for automatic annotation of image data based on deep learning according to claim 1, characterized in that: Use deep learning algorithms to build an automatic image data annotation model, and use reinforcement learning algorithms to build an automatic annotation strategy adjustment model, including the following steps: Collecting a number of historical image data, and preprocessing the number of historical image data to obtain a number of preprocessed historical image data; Based on some pre-processed historical image data, a deep learning algorithm is used to build an automatic image data annotation model; A model adjustment mechanism is set, and according to the model adjustment mechanism, image data is collected to automatically annotate a number of historical performance data of the model, and preprocessing is performed to obtain a number of preprocessed historical performance data; Based on some preprocessed historical performance data, a reinforcement learning algorithm is used to build an automatic labeling strategy adjustment model.

3. The method for automatic annotation of image data based on deep learning according to claim 2, characterized in that: The image data automatic annotation model is constructed based on the FPN-Faster R-CNN-DANN-DBN algorithm, and the image data automatic annotation model includes a graph feature extraction module constructed based on the FPN algorithm, a target area positioning module constructed based on the Faster R-CNN algorithm, a domain adversarial training module constructed based on the DANN algorithm, and an automatic annotation module constructed based on the DBN algorithm. The graph feature extraction module, the target area positioning module, and the automatic annotation module are connected in sequence, the domain adversarial training module is connected to the target area positioning module, and the domain adversarial training module includes a label predictor and a domain classifier connected in sequence.

4. The method for automatic annotation of image data based on deep learning according to claim 3, characterized in that: Based on some pre-processed historical image data, a deep learning algorithm is used to build an automatic image data annotation model, including the following steps: Using the CNN algorithm, a basic network architecture of a graph feature extraction module is constructed; the basic network architecture includes alternately connected convolutional layers and pooling layers; Use the FPN algorithm to build a feature pyramid, and connect the outputs of all convolutional layers to the feature pyramid horizontally in a top-down order to obtain the initial graph feature extraction module; Use the Faster R-CNN algorithm to build the initial target area positioning module, and use the DANN algorithm to build the initial domain adversarial training module; Use the DBN algorithm to build the initial automatic annotation module; Integrate the initial graph feature extraction module, the initial target area positioning module, the initial domain adversarial training module and the initial automatic annotation module to obtain the initial image data automatic annotation model; Combining the loss functions of the initial graph feature extraction module, the initial target region localization module, the initial domain adversarial training module, and the initial automatic labeling module, a comprehensive loss function is obtained; Optimizing initial model parameters of the initial image data automatic annotation model to obtain an optimized image data automatic annotation model; Based on the comprehensive loss function, several preprocessed historical image data are used to optimize the training of the optimized image data automatic annotation model to obtain the final image data automatic annotation model.

5. The method for automatic annotation of image data based on deep learning according to claim 4, characterized in that: Taking minimizing the model error as the optimization goal, a swarm intelligence optimization algorithm is used to optimize the initial model parameters of the initial image data automatic annotation model to obtain an optimized image data automatic annotation model.

6. The method for automatic annotation of image data based on deep learning according to claim 5, characterized in that: The automatic labeling strategy adjustment model is constructed based on the MPO-PPO algorithm, and the automatic labeling strategy adjustment model includes a meta-strategy optimization module constructed based on the MPO algorithm and a reinforcement learning module constructed based on the PPO algorithm. The reinforcement learning module is provided with an intelligent agent, a strategy network and an experience replay pool. The intelligent agent is respectively connected to the strategy network, the experience replay pool and the meta-strategy optimization module, and the meta-strategy optimization module is connected to the experience replay pool.

7. The method for automatic annotation of image data based on deep learning according to claim 6, characterized in that: Based on some pre-processed historical performance data, a reinforcement learning algorithm is used to build an automatic labeling strategy adjustment model, which includes the following steps: Use the FCM clustering algorithm to cluster a number of pre-processed historical performance data to obtain a number of cluster centers and corresponding cluster clusters, and set a historical strategy adjustment category for each cluster cluster; Use the experience replay mechanism to initialize the experience replay pool, build the agent of the reinforcement learning module, and define the state space, action space and reward function of the agent; The historical automatic labeling strategy adjustment problem is used as a simulation environment, the PPO algorithm is used to build a strategy network, and the initial reinforcement learning module is obtained by combining the experience replay pool and the intelligent agent. Set several policy adjustment categories as several sub-scenarios of meta-policy optimization, and use the MPO algorithm to build an initial meta-policy optimization module; According to some pre-processed historical performance data, the initial reinforcement learning module is trained and optimized, and the historical policy network parameters of the policy network of the reinforcement learning module in the training optimization are extracted to obtain the final reinforcement learning module, and generate some historical reinforcement learning experiences; Based on several sub-scenarios and several historical policy network parameters, the initial meta-policy optimization module is trained and optimized to obtain the final meta-policy optimization module, and several historical meta-policy optimization experiences are generated; Several historical meta-strategy optimization experiences and several historical reinforcement learning experiences are stored in the experience replay pool of the final reinforcement learning module, and the final meta-strategy optimization module and the final reinforcement learning module are integrated to obtain an automatic labeling strategy adjustment model.

8. The method for automatic annotation of image data based on deep learning according to claim 7, characterized in that: According to the real-time performance data, the automatic annotation strategy of the automatic annotation strategy adjustment model is adjusted using the automatic annotation strategy adjustment model to obtain an adjusted image data automatic annotation model, including the following steps: Obtaining similarities between the real-time performance data and a number of cluster centers, and using the historical strategy adjustment category of the cluster center with the highest similarity as the real-time strategy adjustment category of the real-time performance data; According to the real-time strategy adjustment category, the corresponding real-time meta-strategy optimization experience is extracted from a number of historical meta-strategy optimization experiences in the experience playback pool of the automatic labeling strategy adjustment model; Based on the experience of real-time meta-strategy optimization, the meta-strategy optimization module is used to initialize the real-time policy network parameters of the policy network of the reinforcement learning module to obtain an updated policy network. Randomly extract a number of real-time reinforcement learning experiences from a number of historical reinforcement learning experiences in the experience replay pool, and update the action space of the reinforcement learning module according to the number of real-time reinforcement learning experiences to obtain an updated action space; According to the real-time performance data, the state space of the reinforcement learning module is updated to obtain an updated state space; Use the agent to control the updated policy network to generate the probability distribution of all possible actions in the action space corresponding to each state in the state space; The possible action with the highest probability distribution in the action space is taken as the execution action of the state, and the execution actions of all states in the state space are integrated to obtain several automatic labeling strategy adjustment actions; According to a number of automatic labeling strategy adjustment actions, the automatic labeling strategy of the automatic labeling strategy adjustment model is adjusted to obtain an adjusted image data automatic labeling model.

9. The method for automatic annotation of image data based on deep learning according to claim 8, characterized in that: Collecting real-time image data, using the adjusted image data automatic annotation model to automatically annotate the real-time image data, and obtaining the automatically annotated real-time image data, includes the following steps: Collecting real-time image data, and preprocessing the real-time image data to obtain preprocessed real-time image data; Using the graph feature extraction module of the adjusted image data automatic annotation model, extract several real-time feature graphs of the preprocessed real-time image data, and perform feature graph fusion to obtain real-time fusion features; Using the adjusted image data to automatically annotate the model’s target region positioning module, perform the target region positioning module based on the real-time fusion features to obtain several real-time target regions; The automatic annotation module of the adjusted image data automatic annotation model is used to automatically annotate several real-time target areas according to the real-time fusion features to obtain the automatically annotated real-time image data.

10. A deep learning-based automatic image data annotation system, used to implement the automatic image data annotation method according to any one of claims 1 to 9, characterized in that: The system comprises a model building unit, a performance data collection unit, a model adjustment unit and an automatic marking unit which are connected in sequence.

Images