Enhanced recognition method, system and equipment for fishery administration monitoring image and storage medium
By employing an image enhancement method based on an encoder-decoder structure and a CBAM attention module, combined with the YOLO model, the problems of poor adaptability of image enhancement and unsatisfactory target detection performance in existing technologies are solved, achieving efficient target recognition in complex low-light environments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-12
- Publication Date
- 2026-03-13
AI Technical Summary
In existing fishery administration monitoring technologies, image enhancement methods are prone to overexposure or underexposure in local areas and amplify noise, resulting in poor adaptability. Furthermore, they are not optimized in conjunction with subsequent advanced vision tasks, which affects the target detection effect.
An image enhancement module with an encoder-decoder structure is combined with a CBAM attention module to extract and reconstruct multi-scale features, which are then input into a pre-trained YOLO model for target recognition. The image enhancement and target detection are optimized by a joint loss function.
It improves the robustness and recognition accuracy of images in complex low-light environments, enhances the detailed features of key targets, suppresses noise interference, and improves the environmental adaptability and recognition accuracy of the fishery administration monitoring system.
Smart Images
Figure CN121661416A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of image processing technology, and in particular relates to a method, system, device and storage medium for enhancing and recognizing fishery administration monitoring images. Background Technology
[0002] Fisheries monitoring refers to the use of modern technology by fisheries supervision and management departments to monitor and manage fisheries activities in a real-time, dynamic, and intelligent manner in order to protect fishery resources, maintain fisheries order, combat illegal fishing, and promote the sustainable development of fisheries.
[0003] In existing technologies, water areas are monitored in real time using video cameras, and image enhancement methods are used for image processing. For example, enhancement methods based on traditional image processing, such as histogram equalization (HE) and homomorphic filtering, are used; or deep learning-based image enhancement methods are used to enhance low-light performance and reduce noise in the monitored images.
[0004] The existing technology has the following technical problems: 1. Existing technologies involve global adjustments, which can easily lead to overexposure or underexposure in local areas and amplify noise, resulting in limited effectiveness and poor adaptability.
[0005] 2. Existing technologies aim to improve the subjective visual quality of images without co-optimizing with subsequent advanced visual tasks (such as object detection). Simply enhancing the visual effect of an image may introduce artifacts or noise that are detrimental to object detection, or excessively enhance the background while weakening key target features, creating a target gap between "enhancement" and "recognition". Summary of the Invention
[0006] This invention provides a method, system, device, and storage medium for enhancing and recognizing images in fishery administration monitoring, aiming to solve the technical problems in the prior art, such as easy overexposure or underexposure in local areas, amplification of noise, limited effectiveness, poor adaptability, and lack of co-optimization with subsequent advanced vision tasks.
[0007] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for enhancing and recognizing fishery administration monitoring images, comprising: Obtain fishery administration monitoring images; The fishery administration monitoring images are subjected to image enhancement processing to obtain enhanced images; The enhanced image is input into a pre-trained target recognition model for target recognition, and the recognition result is obtained.
[0008] Furthermore, the above also includes: Determine whether an alarm needs to be triggered based on the identification results: If no alarm is needed, then end the process. If an alarm is required, an alarm notification will be generated.
[0009] Furthermore, the aforementioned alarm notification includes target identification and the target's location information.
[0010] Furthermore, the above-mentioned image enhancement processing of the fishery administration monitoring image yields the enhanced image as follows: The multi-scale features of the fishery monitoring images are extracted using an encoder composed of multiple convolutional layers and pooling layers to obtain a multi-scale feature image; The CBAM attention module is used to calculate feature weights for the channel dimension and spatial dimension respectively, to obtain channel feature weights and spatial feature weights; wherein, the CBAM attention module is set in each hop connection between the encoder and the decoder; A decoder consisting of multiple deconvolutional and pooling layers is used to reconstruct multi-scale feature images by combining the channel feature weights and spatial feature weights, thereby obtaining an enhanced image.
[0011] Furthermore, the enhanced image is input into the pre-trained target recognition model for target recognition, and the recognition result is as follows: We constructed a YOLO model that deeply integrates channel attention and spatial attention mechanisms, and obtained the training dataset. The YOLO model is trained using the training dataset to obtain the target recognition model; The target recognition model is integrated into the target recognition module to perform real-time target detection on the enhanced image and obtain the recognition result; wherein, the recognition result includes a bounding box with confidence, category label and structured event information.
[0012] Furthermore, the above-mentioned acquisition of the training dataset specifically involves: Obtain a set of historical fishery administration monitoring images, and annotate the set of historical fishery administration monitoring images to obtain an annotated image set; The labeled image set is subjected to low-light augmentation processing to obtain a training dataset; wherein, the low-light augmentation processing includes: random brightness and contrast adjustment, multi-type noise injection, and motion blur simulation.
[0013] Furthermore, the above also includes: The joint loss function is constructed as shown in the following equation:
[0014] Among them, L pixel This represents the L1 loss between the enhanced image and the fisheries monitoring image; L perceptual This represents the feature loss between the enhanced image and the fisheries monitoring image; L detect λ represents the detection loss of the YOLO model for predicting augmented images; λ1, λ2, and λ3 represent the weight hyperparameters, respectively. During model training, the parameters of the YOLO model are dynamically updated using the backpropagation method.
[0015] Secondly, to solve the above-mentioned technical problems, the present invention also provides a fishery administration monitoring image enhancement and recognition system, comprising: The data acquisition module is used to acquire images from fisheries monitoring systems. The low-light image enhancement module is used to perform image enhancement processing on the fishery administration monitoring image to obtain an enhanced image; The target recognition module is used to input the enhanced image into a pre-trained target recognition model for target recognition and obtain the recognition result.
[0016] Thirdly, in order to solve the above-mentioned technical problems, the present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the fishery administration monitoring image enhancement and recognition method of the present application.
[0017] Fourthly, in order to solve the above-mentioned technical problems, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the fishery administration monitoring image enhancement and recognition method of the present application.
[0018] Compared with the prior art, the present invention has the following advantages: 1. This invention first processes the input low-light monitoring image using a low-light image enhancement module. This module employs an encoder-decoder structure and incorporates an attention mechanism to effectively preserve and enhance the detailed features of targets such as vessels and personnel while suppressing noise, thereby improving the overall brightness and contrast of the image. Subsequently, the enhanced image is input into a pre-trained target detection neural network to achieve accurate identification and location of specific targets such as fishing boats, fishing activities, and personnel. This invention achieves end-to-end collaborative optimization of image enhancement and target recognition, ensuring that the image enhancement process directly serves to improve recognition accuracy rather than simply pursuing visual appeal. This significantly enhances the robustness and practicality of the fisheries monitoring system in complex low-light environments.
[0019] 2. This invention improves the recognition accuracy in low light by using end-to-end joint training and a joint optimization objective that includes detection loss, so that the image output by the image enhancement module contains the feature information most conducive to the operation of the target detector.
[0020] 3. The attention mechanism introduced in this invention enables the model to automatically focus on key target areas in complex low-light scenes, suppress interference from irrelevant backgrounds and noise, and enhance the system's environmental adaptability.
[0021] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description
[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 A flowchart illustrating a fisheries administration monitoring image enhancement and recognition method according to an embodiment of the present invention is shown. Figure 2 A schematic diagram of a fisheries monitoring image according to an embodiment of the present invention is shown; Figure 3 An enhanced image schematic diagram according to an embodiment of the present invention is shown; Figure 4 A schematic diagram of the recognition result image according to an embodiment of the present invention is shown; Figure 5 A schematic flowchart of the enhancement process according to an embodiment of the present invention is shown; Figure 6 A schematic diagram of the structure of a fishery administration monitoring image enhancement and recognition method according to an embodiment of the present invention is shown; Figure 7 A schematic diagram of an electronic device structure according to an embodiment of the present invention is shown. Detailed Implementation
[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0025] Figure 1 A flowchart illustrating a fisheries monitoring image enhancement and recognition method according to an embodiment of the present invention is shown, as follows: Figure 1 As shown, an embodiment of the present invention provides a fishery administration monitoring image enhancement and recognition method, comprising: Obtain fishery administration monitoring images; The fishery administration monitoring images are subjected to image enhancement processing to obtain enhanced images; The enhanced image is input into a pre-trained target recognition model for target recognition, and the recognition result is obtained.
[0026] like Figure 2 , Figure 3 ,and Figure 4 The images shown are the original fisheries monitoring image, the enhanced image, and the recognition result image, respectively, from this embodiment. Figure 3 As can be seen, the enhanced image is bright, the outlines of targets such as ships are clear, and noise control is good. Figure 3 Based on this, the fishing vessel targets were accurately outlined with rectangular boxes, and their categories and confidence levels were labeled, forming... Figure 4 .
[0027] Optional, also includes: Determine whether an alarm needs to be triggered based on the identification results: If no alarm is needed, then end the process. If an alarm is required, an alarm notification will be generated.
[0028] Optionally, the alarm notification includes target identification and target location information.
[0029] Optionally, the fishery administration monitoring image is subjected to image enhancement processing to obtain an enhanced image, specifically as follows: The multi-scale features of the fishery monitoring images are extracted using an encoder composed of multiple convolutional layers and pooling layers to obtain a multi-scale feature image; The CBAM attention module is used to calculate feature weights for the channel dimension and spatial dimension respectively, to obtain channel feature weights and spatial feature weights; wherein, the CBAM attention module is set in each hop connection between the encoder and the decoder; A decoder consisting of multiple deconvolutional and pooling layers is used to reconstruct multi-scale feature images by combining the channel feature weights and spatial feature weights, thereby obtaining an enhanced image.
[0030] In this embodiment, as Figure 5 As shown, the encoder's operation steps include: Convolutional layer operation: The input image is convolved sequentially through multiple convolutional layers. Each convolutional layer is followed by a non-linear activation function (such as ReLU) to extract local features.
[0031] Pooling layer downsampling: A pooling layer (usually max pooling) is applied after every few convolutional layers to reduce the size of the feature map while retaining the most important information, gradually forming a multi-scale feature representation.
[0032] The steps for operating the CBAM attention module include: Channel-dimensional attention calculation: First, the importance weight of features is calculated in the channel dimension, emphasizing the feature channels that are more important to the final task (i.e., the identification and enhancement of the target region).
[0033] Spatial Dimension Attention Calculation: Next, the importance weights of each location in the feature map are calculated in the spatial dimension to highlight regions that may contain important targets.
[0034] Weighted fusion: The feature maps processed by channel and spatial attention mechanisms are weighted and fused with the corresponding encoder output to generate a feature representation with attention guidance.
[0035] The decoder's operation steps include: Deconvolution: Deconvolution layers are used to restore the spatial resolution of the feature map, gradually approaching the size of the original image.
[0036] Combining skip connection information: At each decoding stage, the feature map processed by CBAM from the corresponding encoder level is combined with the feature map of the current decoder level to help restore details and reduce artifacts.
[0037] Layer-by-layer reconstruction: As the number of layers increases, a clearer and more detailed enhanced image is gradually reconstructed.
[0038] Optionally, the enhanced image is input into a pre-trained target recognition model for target recognition, and the recognition result is as follows: We constructed a YOLO model that deeply integrates channel attention and spatial attention mechanisms, and obtained the training dataset. The YOLO model is trained using the training dataset to obtain the target recognition model; The target recognition model is integrated into the target recognition module to perform real-time target detection on the enhanced image and obtain the recognition result; wherein, the recognition result includes a bounding box with confidence, category label and structured event information.
[0039] In this embodiment, the YOLO series (such as YOLOv5, YOLOv8, or YOLO-NAS) is selected as the basic target detection framework because it strikes a good balance between detection speed and accuracy, making it suitable for the real-time requirements of fishery administration monitoring. At key locations in the YOLO backbone or feature pyramid network, channel attention and spatial attention mechanisms (such as CBAM or SE+SA combinations) are deeply integrated. Through this mechanism, the model can adaptively focus on key target areas such as fishing boats, fishing gear, and personnel under complex backgrounds, low contrast, or high noise conditions, improving the detection rate of small and blurred targets.
[0040] Optionally, the training dataset is obtained as follows: Obtain a set of historical fishery administration monitoring images, and annotate the set of historical fishery administration monitoring images to obtain an annotated image set; The labeled image set is subjected to low-light augmentation processing to obtain a training dataset; wherein, the low-light augmentation processing includes: random brightness and contrast adjustment, multi-type noise injection, and motion blur simulation.
[0041] In this embodiment, the historical fisheries monitoring image set is derived from real historical data collected during previous project implementations, covering video streams and images at night, dawn, dusk, and in inclement weather. The acquisition equipment uses cameras of different grades (ordinary visible light, low-light starlight, near-infrared) to study the impact of different imaging devices. Scenes and objectives: ensuring coverage of different water bodies such as rivers, lakes, and reservoirs; objectives include various fishing vessels (trawlers, gillnetters, angling boats, etc.), various fishing gear (fish traps, floating nets, electrofishing equipment, etc.), various fish species (carp, grass carp, crucian carp, etc.), and various fishing behaviors (electrofishing, net fishing, angling, etc.).
[0042] During the annotation process, professional tools such as LabelImg and CVAT are used.
[0043] Labeling Standards: Use bounding boxes for labeling. Define a clear category system, such as "fishing boat," "fisherman," "floating fishing net," and "angling boat." Obscure or difficult-to-identify targets should also be labeled and marked as "difficult examples" for subsequent analysis.
[0044] After labeling, the training set, validation set, and test set are randomly divided in a 6:2:2 ratio to ensure a balanced distribution of each category and scenario.
[0045] In this embodiment, random brightness and contrast adjustment is used to simulate the imaging effect under different light intensities; multiple types of noise injection are used, such as Gaussian noise and salt-and-pepper noise, to simulate sensor noise or transmission interference; motion blur simulation is used to simulate the blur caused by the movement of ships or cameras through convolution kernels.
[0046] Optional, also includes: The joint loss function is constructed as shown in the following equation:
[0047] Among them, L pixel This represents the L1 loss between the enhanced image and the fisheries monitoring image; L perceptual This represents the feature loss between the enhanced image and the fisheries monitoring image; L detect λ represents the detection loss of the YOLO model for predicting augmented images; λ1, λ2, and λ3 represent the weight hyperparameters, respectively. During model training, the parameters of the YOLO model are dynamically updated using the backpropagation method.
[0048] In this embodiment, the feature loss is calculated using the ReLU3_3 layer of the pre-trained VGG-16 network, and the detection loss includes box regression loss, confidence loss, and classification loss. Each loss value is obtained through forward propagation, and then the total loss value L is calculated. total Perform backpropagation to calculate the gradient; update the YOLO model parameters according to the training strategy to iteratively optimize the YOLO model until it converges, thus obtaining the target recognition model.
[0049] Based on and Figure 1 Using the same principle as the method shown, this embodiment of the invention also provides a fishery administration monitoring image enhancement and recognition system, such as... Figure 6 As shown, it includes: The data acquisition module is used to acquire images from fisheries monitoring systems. The low-light image enhancement module is used to perform image enhancement processing on the fishery administration monitoring image to obtain an enhanced image; The target recognition module is used to input the enhanced image into a pre-trained target recognition model for target recognition and obtain the recognition result.
[0050] In this embodiment, the low-light image enhancement module can be replaced by a Transformer-based enhancement network (such as SwinIR), which improves feature extraction capabilities through a multi-head attention mechanism, but still needs to be trained end-to-end with the target recognition module; the target recognition module can be replaced by an SSD or Faster R-CNN network to adapt to fishery monitoring scenarios with different accuracy and speed requirements.
[0051] The fishery administration monitoring image enhancement and recognition system of this invention can execute the fishery administration monitoring image enhancement and recognition method provided in this invention. The implementation principle is similar. The actions performed by each module and unit in the fishery administration monitoring image enhancement and recognition system in each embodiment of this invention correspond to the steps in the fishery administration monitoring image enhancement and recognition method in each embodiment of this invention. For detailed functional descriptions of each module of the fishery administration monitoring image enhancement and recognition system, please refer to the descriptions in the corresponding fishery administration monitoring image enhancement and recognition methods shown above. They will not be repeated here.
[0052] The aforementioned fishery administration monitoring image enhancement and recognition system can be a computer program (including program code) running on a computer device, such as an application software; the application software can be used to execute the corresponding steps in the method provided in the embodiments of the present invention.
[0053] The modules described in the embodiments of the present invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.
[0054] Based on the same principles as the methods shown in the embodiments of the present invention, the embodiments of the present invention also provide an electronic device, which may include, but is not limited to: a processor and a memory; the memory for storing computer programs; and the processor for executing the methods shown in any embodiment of the present invention by invoking the computer programs.
[0055] In one alternative embodiment, an electronic device is provided, such as Figure 7 As shown, Figure 7 The illustrated electronic device includes a processor and a memory. The processor and memory are connected, for example, via a bus. Optionally, the electronic device may also include a transceiver, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver is not limited to one unit, and the structure of this electronic device does not constitute a limitation on the embodiments of the present invention.
[0056] The memory stores application code (computer program) that executes the present invention, and its execution is controlled by a processor. The processor executes the application code stored in the memory to implement the content shown in the foregoing method embodiments.
[0057] Among these, electronic devices can also be terminal devices. Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.
[0058] This invention provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.
[0059] According to another aspect of the present invention, a computer program product or computer program is also provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various embodiments described above.
[0060] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.
[0061] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.
Claims
1. A method for enhancing and recognizing images from fishery administration monitoring, characterized in that, The method includes: Obtain fishery administration monitoring images; The fishery administration monitoring images are subjected to image enhancement processing to obtain enhanced images; The enhanced image is input into a pre-trained target recognition model for target recognition, and the recognition result is obtained.
2. The fishery administration monitoring image enhancement and recognition method according to claim 1, characterized in that, Also includes: Determine whether an alarm needs to be triggered based on the identification results: If no alarm is needed, then end the process. If an alarm is required, an alarm notification will be generated.
3. The fishery administration monitoring image enhancement and recognition method according to claim 2, characterized in that, The alarm notification includes target identification and target location information.
4. The fishery administration monitoring image enhancement and recognition method according to claim 1, characterized in that, The fishery administration monitoring image is subjected to image enhancement processing to obtain the enhanced image as follows: The multi-scale features of the fishery monitoring images are extracted using an encoder composed of multiple convolutional layers and pooling layers to obtain a multi-scale feature image; The CBAM attention module is used to calculate feature weights for the channel dimension and spatial dimension respectively, to obtain channel feature weights and spatial feature weights; wherein, the CBAM attention module is set in each hop connection between the encoder and the decoder; A decoder consisting of multiple deconvolutional and pooling layers is used to reconstruct multi-scale feature images by combining the channel feature weights and spatial feature weights, thereby obtaining an enhanced image.
5. The fishery administration monitoring image enhancement and recognition method according to claim 1, characterized in that, The enhanced image is input into a pre-trained target recognition model for target recognition, and the recognition result is as follows: We constructed a YOLO model that deeply integrates channel attention and spatial attention mechanisms, and obtained the training dataset. The YOLO model is trained using the training dataset to obtain the target recognition model; The target recognition model is integrated into the target recognition module to perform real-time target detection on the enhanced image and obtain the recognition result; wherein, the recognition result includes a bounding box with confidence, category label and structured event information.
6. The fishery administration monitoring image enhancement and recognition method according to claim 5, characterized in that, The specific steps for obtaining the training dataset are as follows: Obtain a set of historical fishery administration monitoring images, and annotate the set of historical fishery administration monitoring images to obtain an annotated image set; The labeled image set is subjected to low-light augmentation processing to obtain a training dataset; wherein, the low-light augmentation processing includes: random brightness and contrast adjustment, multi-type noise injection, and motion blur simulation.
7. The fishery administration monitoring image enhancement and recognition method according to claim 5, characterized in that, Also includes: The joint loss function is constructed as shown in the following equation: Among them, L pixel This represents the L1 loss between the enhanced image and the fisheries monitoring image; L perceptual This represents the feature loss between the enhanced image and the fisheries monitoring image; L detect λ represents the detection loss of the YOLO model for predicting augmented images; λ1, λ2, and λ3 represent the weight hyperparameters, respectively. During model training, the parameters of the YOLO model are dynamically updated using the backpropagation method.
8. A fisheries administration monitoring image enhancement and recognition system, characterized in that, include: The data acquisition module is used to acquire images from fisheries monitoring systems. The low-light image enhancement module is used to perform image enhancement processing on the fishery administration monitoring image to obtain an enhanced image; The target recognition module is used to input the enhanced image into a pre-trained target recognition model for target recognition and obtain the recognition result.
9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method of any one of claims 1-7.
10. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the method of any one of claims 1-7.