AI identification early warning method for updating surrounding environment in real time and improving identification accuracy

Abstract

The invention discloses an AI identification early warning method for updating surrounding environment in real time and improving identification accuracy, and belongs to the technical field of artificial intelligence and image identification early warning. The method comprises the following steps: 1, an environment change perception module: 1, physical environment perception: analyzing the change (such as tree growth, grassland coverage, land form change and the like) of the surrounding environment through an image recognition algorithm, and detecting the physical change of the surrounding environment by comparing a current image with a historical image, for example, calculating the relative position of a tree and a power transmission line; 2, sensor data acquisition: in addition to image data, the system can further analyze the change of the surrounding environment in combination with an environment sensor and a position sensor, the real-time performance is high, the change of the surrounding environment of the power transmission line can be quickly captured, real-time detection and early warning can be carried out on possible dangerous conditions, the delay is reduced, and the safety of the power transmission line is ensured; and the emergency response capability of the system is improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of artificial intelligence and image recognition warning technology, and in particular to an AI recognition warning method for improving recognition accuracy by real-time updating the surrounding environment. Background Art

[0002] The safety monitoring of transmission lines is an important part of the power system. Current monitoring systems usually analyze and identify based on static models or historical data. However, the environment around transmission lines changes significantly over time, such as seasonal vegetation growth, construction activities, newly built buildings or roads, and sudden changes in climate conditions. These changes may lead to a decrease in the recognition accuracy of existing monitoring models, or even cause false alarms or missed alarms. In addition, the update cycle of big data models is relatively long, and the data of the surrounding environment cannot be updated in real time, making it difficult to identify accurate data of the surrounding environment and difficult to judge whether there is danger in the actual surrounding environment, which cannot meet the requirements of real-time danger detection.

[0003] In view of the above problems, the present invention is improved. Summary of the Invention

[0004] The present invention provides an AI recognition warning method for improving recognition accuracy by real-time updating the surrounding environment, which solves the above problems existing in the prior art during use.

[0005] The technical solution of the present invention is realized as follows: An AI recognition warning method for improving recognition accuracy by real-time updating the surrounding environment, and the recognition and capture are as follows:

[0006] I. Environmental change perception module:

[0007] 1. Physical environment perception: Analyze the changes in the surrounding environment (such as tree growth, grass coverage, landform changes, etc.) through image recognition algorithms. By comparing the current image with the historical image, detect the physical changes in the surrounding environment. For example, by calculating the relative position between the tree and the transmission line, determine whether the tree may affect the safety of the transmission line;

[0008] 2. Sensor data acquisition: In addition to image data, the system can also combine environmental sensors (such as temperature and humidity, air pressure sensors) and position sensors (such as GPS, lidar, etc.) to further analyze the changes in the surrounding environment;

[0009] II. Data acquisition and preprocessing:

[0010] 1. Image data acquisition: The camera regularly acquires environmental images to capture dynamic environmental information such as tree growth and grass changes;

[0011] 2. Environmental change recognition: Through image processing and comparative analysis, identify environmental changes in the image, such as whether the trees have grown taller or if there are new obstacles (e.g., new buildings, changes in grass coverage, etc.); 3. Incremental data processing: Based on the detected environmental changes, perform incremental preprocessing and enhancement on the image data in real-time to ensure that the subsequent training input data contains the characteristics of the current environment;

[0012] III. Initial YOLO model deployment:

[0013] 1. Deploy the YOLO model trained based on the existing environment (such as the initial state of trees, grass, etc.) on the terminal device. This model can identify common dangerous objects (such as tree branches, flying birds, etc.) and give preliminary danger warnings;

[0014] IV. Environmental change detection and incremental training mechanism:

[0015] 1. Environmental change detection: By comparing with historical image data, analyze the changes in the surrounding environment in real-time. If the environment changes significantly (such as trees growing taller, changes in grass coverage, etc.), the system will trigger the incremental training process;

[0016] 2. Incremental training process: Through the incremental training algorithm, fine-tune the YOLO model based on the latest environmental data. The focus of incremental training is partial update of the model weights rather than retraining from scratch to avoid wasting computing resources;

[0017] 2.1 For objects related to environmental changes (such as tree occlusion, wire obstruction, etc.), the system will make special adjustments to the model according to the change data of the actual environment;

[0018] V. Adaptive model update and feedback mechanism:

[0019] 1. Real-time feedback and adaptive adjustment: Based on the results of incremental training triggered by environmental changes, the system will evaluate the effect of model update in real-time. If the new model version can better identify the risks brought by environmental changes (such as occlusions, obstacles, etc.), the system will automatically switch to the new version;

[0020] 2. Model evaluation and rollback: When the system detects that the performance of the new model is not as expected (such as an increase in false alarms and missed alarms), it can quickly roll back to the previous stable version through the feedback mechanism to ensure the stability of the system;

[0021] VI. Fusion of environmental data and image data:

[0022] 1. Environmental data fusion: In addition to image data, the system combines environmental data (such as humidity, wind speed, temperature, etc.) to analyze and judge the impact of environmental changes. For example, high-humidity weather may cause the growth rate of trees and grass to accelerate. The system can combine this data to predict changes in advance and make corresponding adjustments to the model;

[0023] 2. Combined input of real-time images and environmental data: During the incremental training process, the combined input of environmental data and image data will help the model more accurately perceive the risk points related to environmental changes;

[0024] VII. Distributed update and synchronization:

[0025] 1. Distributed training and update: When multiple terminal devices share the same area, the updated model will be synchronized to other devices through the cloud or edge computing network to ensure that all devices use the same updated model when dealing with the same environmental changes.

[0026] 2. Multi-terminal collaboration: When multiple terminal devices work together, they can share the model after incremental training and perform dynamic optimization according to environmental changes in different regions.

[0027] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: In step six, dynamic environmental change perception: Through image analysis and sensor data, the physical changes in the surrounding environment (such as tree growth, grass coverage change, etc.) are detected in real time, and the model is adjusted according to these changes to enhance the recognition ability of the system in a dynamic environment.

[0028] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: In step six, incremental training and environmental adaptation: Through the incremental training mechanism, not only new objects are trained, but also fine-tuning is performed for environmental changes (such as changes in tree occlusion of power lines, changes in grasslands, etc.), so as to ensure the adaptability of the model to the new environment.

[0029] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: In step six, real-time model update and feedback mechanism: According to environmental changes, real-time updates and model fine-tuning are triggered to optimize the recognition effect of the model in a complex environment and avoid false alarms and missed alarms caused by environmental changes.

[0030] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: In step six, environmental perception and automated feedback mechanism: Through environmental data (such as image change analysis, sensor feedback, etc.), it is automatically judged whether there are environmental changes, triggering incremental training and automatically adjusting the model to ensure that the model always adapts to the latest environmental state.

[0031] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: in the environmental change detection in step 4, if it is found that the growth of trees blocks the transmission line, the system will record the new growth situation of the trees and update the model to recognize the new tree form or relative position. If the grass coverage changes and affects the safety monitoring of the transmission line, the system will also re-label and train the relevant data.

[0032] An AI recognition device for real-time updating of the surrounding environment to improve the recognition accuracy, including a bracket, a pan-tilt is fixedly installed at the top of the bracket, an environmental change perception module is fixedly connected to one side of the pan-tilt, a control terminal is fixedly installed at the top of the pan-tilt, a supplementary light module and a picture capture module are respectively fixedly installed on both sides of the control terminal, a support is fixedly installed at the bottom of the bracket, and radars are fixedly installed at the bottoms of both sides of the support.

[0033] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: a cross bar is detachably installed inside the bracket, a fitting frame is fixedly installed at one end of the cross bar away from the bracket, and a mounting frame is fixedly installed inside the fitting frame.

[0034] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: a plurality of mounting sleeves are sleeved on the surface of the cross bar, and a plurality of telescopic rods are movably installed on the surface of one of the mounting sleeves.

[0035] For the AI recognition and warning method for real-time updating of the surrounding environment to improve the recognition accuracy as described above in the present invention, further: the telescopic rod includes a fixed rod movably installed on the mounting sleeve, an adjusting screw rod is threadedly installed at one end of the fixed rod, and an extension rod is fixedly connected to the end of the adjusting screw rod away from the fixed rod.

[0036] In summary, the beneficial effects of the present invention are as follows:

[0037] 1. High real-time performance, rapid response to risks. Through the real-time updated big data model, the system can quickly capture the changes in the surrounding environment of the transmission line, instantaneously detect and warn of possible dangerous situations, reduce delays, and enhance the emergency response ability of the system.

[0038] Strong dynamic adaptability. The big data model can be adaptively updated according to the real-time collected environmental data (such as meteorological information, dynamic changes of surrounding obstacles, etc.), ensuring that the detection model always matches the current environmental conditions and reducing the false alarm rate and missed alarm rate.

[0039] Precisely identify potential threats. By adopting an intelligent algorithm based on big data, multi-dimensional data (such as distance, speed, direction, object features, etc.) can be comprehensively analyzed to accurately judge the type and severity of threats, so as to formulate more effective countermeasures.

[0040] 2. Through the settings of the mounting frame, adapter frame and cross bar, the mounting frame is used to be installed on the electric pole, so that the entire recognition system can be assembled at the required position, thereby achieving the recognition accuracy of the surrounding environment of the transmission line, protecting the transmission line from being damaged, ensuring the normal operation of the transmission line, and providing convenience for the user's work and use.

[0041] 3. Through the setting of the mounting sleeve, a plurality of mounting sleeves are installed on the cross bar. In this way, other components can be installed on other mounting sleeves according to actual usage requirements to ensure the normal operation of the entire device, ensure that the entire device realizes its due functions, ensure the normal operation of the device, and provide convenience for the user's work and use.

[0042] 4. Through the setting of the telescopic rod, the telescopic rod can be adjusted in length. When in use, rotate the extension rod to drive the adjustment screw rod to rotate. In this way, the adjustment screw rod will turn out from the fixed rod, thereby extending the length of the telescopic rod to meet more usage requirements and providing convenience for the user's work and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0044] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0045] Figure 2 It is a schematic perspective view of a partial structure of the present invention;

[0046] Figure 3 It is a schematic diagram of the overall structure of;

[0047] Figure 4 It is a schematic perspective view of the telescopic rod.

[0048] In the figure: 1. Mounting frame, 2. Adapter frame, 3. Cross bar, 4. Bracket, 5. Pan-tilt, 6. Ambient change perception module, 7. Control terminal, 8. Supplementary light module, 9. Image capture module, 10. Bracket, 11. Radar, 12. Mounting sleeve, 13. Telescopic rod, 131. Fixed rod, 132. Adjustment screw rod, 133. Extension rod. Detailed implementation manners

[0049] The following will combine the accompanying drawings in the embodiments of the present invention Figures 1-4 to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment

[0051] An AI recognition and warning method for real-time updating the surrounding environment to improve the recognition accuracy, and the recognition and capture process is as follows:

[0052] I. Environmental change perception module:

[0053] 1. Physical environment perception: Analyze the changes in the surrounding environment (such as tree growth, grass coverage, landform changes, etc.) through image recognition algorithms. By comparing the current image with the historical image, detect the physical changes in the surrounding environment. For example, by calculating the relative position between the tree and the power transmission line, determine whether the tree may affect the safety of the power transmission line;

[0054] 2. Sensor data collection: In addition to image data, the system can also combine environmental sensors (such as temperature and humidity, air pressure sensors) and position sensors (such as GPS, lidar, etc.) to further analyze the changes in the surrounding environment;

[0055] II. Data collection and preprocessing:

[0056] 1. Image data collection: The camera regularly collects environmental images to capture dynamic environmental information such as tree growth and grass changes;

[0057] 2. Environmental change recognition: Through image processing and comparative analysis, identify the environmental changes in the image, such as whether the tree has grown taller or whether there are new obstacles (such as new buildings, grass coverage changes, etc.);

[0058] 3. Incremental data processing: According to the detected environmental changes, perform incremental preprocessing and enhancement on the image data in real time to ensure that the subsequent training input data contains the characteristics of the current environment;

[0059] III. Initial YOLO model deployment:

[0060] 1. Deploy the YOLO

[0061] model trained based on the existing environment (such as the initial state of trees, grass, etc.) on the terminal device. This model can identify common dangerous objects (such as tree branches, flying birds, etc.) and perform preliminary danger warnings;

[0062] IV. Environmental Change Detection and Incremental Training Mechanism:

[0063] 1. Environmental change detection: By comparing with historical image data, the system analyzes the changes in the surrounding environment in real time. If significant environmental changes occur (such as trees growing taller, changes in grassland coverage, etc.), the system will trigger the incremental training process.

[0064] 2. Incremental training process: Through the incremental training algorithm, the YOLO model is fine-tuned based on the latest environmental data. The focus of incremental training is partial update of the model weights rather than retraining from scratch to avoid wasting computing resources.

[0065] 2.1 For objects related to environmental changes (such as tree occlusion, wire obstruction, etc.), the system will make specific adjustments to the model according to the change data of the actual environment.

[0066] V. Adaptive Model Update and Feedback Mechanism:

[0067] 1. Real-time feedback and adaptive adjustment: According to the results of incremental training triggered by environmental changes, the system evaluates the effect of model updates in real time. If the new model version can better identify risks brought by environmental changes (such as occlusions, obstacles, etc.), the system will automatically switch to the new version.

[0068] 2. Model evaluation and rollback: When the system detects that the performance of the new model is not as expected (such as an increase in false alarms and missed detections), it can quickly roll back to the previous stable version through the feedback mechanism to ensure the stability of the system.

[0069] VI. Fusion of Environmental Data and Image Data:

[0070] 1. Environmental data fusion: In addition to image data, the system combines environmental data (such as humidity, wind speed, temperature, etc.) to analyze and judge the impact of environmental changes. For example, high humidity weather may cause the growth rate of trees and grasslands to accelerate. The system can combine this data to predict changes in advance and make corresponding adjustments to the model.

[0071] 2. Combined input of real-time images and environmental data: During the incremental training process, the fused input of environmental data and image data will help the model more accurately perceive risk points related to environmental changes.

[0072] VII. Distributed Update and Synchronization:

[0073] 1. Distributed training and update: When multiple terminal devices share the same area, the updated model will be synchronized to other devices through the cloud or edge computing network to ensure that all devices use the same updated model when dealing with the same environmental changes.

[0074] 2. Multi - terminal collaboration: When multiple terminal devices work collaboratively, they can share the model after incremental training and perform dynamic optimization according to environmental changes in different regions.

[0075] In step six mentioned above, dynamic environmental change perception: Through image analysis and sensor data, physically detect real - time changes in the surrounding environment (such as tree growth, grass coverage changes, etc.), and adjust the model according to these changes to enhance the system's recognition ability in a dynamic environment.

[0076] In step six mentioned above, incremental training and environmental adaptation: Through the incremental training mechanism, not only train new objects but also fine - tune for environmental changes (such as changes in tree occlusion of power lines, changes in grasslands, etc.), so as to ensure the adaptability of the model to the new environment.

[0077] In step six mentioned above, real - time model update and feedback mechanism: Trigger real - time update and model fine - tuning according to environmental changes, optimize the model's recognition effect in a complex environment, and avoid false alarms and missed detections caused by environmental changes.

[0078] In step six mentioned above, environmental perception and automated feedback mechanism: Automatically judge whether there are environmental changes through environmental data (such as image change analysis, sensor feedback, etc.), trigger incremental training and automatically adjust the model to ensure that the model always adapts to the latest environmental state.

[0079] In the 1. Environmental change detection of step four, if it is found that tree growth occludes the power line, the system will record the new growth situation of the tree and update the model to identify the new tree form or relative position. If the grass coverage changes and affects the safety monitoring of the power line, the system will also re - label and train the relevant data.

[0080] An AI recognition device for improving recognition accuracy by real - time updating the surrounding environment, including a bracket 4. A pan - tilt 5 is fixedly installed at the top of the bracket 4. An environmental change perception module 6 is fixedly connected to one side of the pan - tilt 5. A control terminal 7 is fixedly installed at the top of the pan - tilt 5. A supplementary lighting module 8 and an image capture module 9 are respectively fixedly installed on both sides of the control terminal 7. A support 10 is fixedly installed at the bottom of the bracket 4. Radars 11 are fixedly installed at the bottoms of both sides of the support 10.

[0081] The specific usage process is as follows: First, the image capture module 9 is used to capture the images of things. The fill light module 8 can supplement the light source to ensure that the image capture module 9 captures things more clearly. Then, the captured data is transmitted to the control terminal 7 for processing, analysis, and recognition, or is processed, analyzed, and recognized by the central control device. And continuously capture and analyze the states of things at different times, and then the various modules in the system process and analyze them respectively, and perform analysis and recognition at the upload terminal, such as the environmental change perception module 6, etc. For the specific AI recognition method of this solution, please refer to it. Among them, there is also a pan-tilt 5 used as a data transfer station to ensure that there is enough storage to update the images in real time, and it is also used to rotate the environmental change perception module 6, control terminal 7, fill light module 8, and image capture module 9 in real time to ensure that the recognition system can capture the images of things in all directions. In addition, the system is also equipped with a radar 11 for recognizing things at a long distance, such as the situation where a bird flies from afar, etc. In this way, the analysis and recognition of things are carried out from multiple aspects and angles to improve the accuracy of environmental recognition and provide convenience for users' work and use.

[0082] A cross bar 3 is detachably installed inside the bracket 4. One end of the cross bar 3 away from the bracket 4 is fixedly installed with an adapter bracket 2, and an installation bracket 1 is fixedly installed inside the adapter bracket 2.

[0083] Specifically, due to the settings of the installation bracket 1, adapter bracket 2, and cross bar 3, the installation bracket 1 is used to be installed on the electric pole, so that the entire recognition system can be assembled at the required position, thereby realizing the recognition accuracy of the surrounding environment of the transmission line, protecting the transmission line from being damaged, ensuring the normal operation of the transmission line, and providing convenience for users' work and use.

[0084] A plurality of mounting sleeves 12 are sleeved on the surface of the cross bar 3, and a plurality of telescopic rods 13 are movably installed on the surface of one of the mounting sleeves 12.

[0085] Specifically, due to the setting of the mounting sleeve 12, a plurality of mounting sleeves 12 are installed on the cross bar 3. In this way, other components can be installed on other mounting sleeves 12 according to actual usage requirements to ensure the normal operation of the entire device, ensure that the entire device realizes its due functions, ensure the normal operation of the device, and provide convenience for users' work and use.

[0086] The telescopic rod 13 includes a fixed rod 131 movably installed on the mounting sleeve 12. One end of the fixed rod 131 is threadedly installed with an adjusting screw rod 132, and the end of the adjusting screw rod 132 away from the fixed rod 131 is fixedly connected with an extension rod 133.

[0087] Specifically, the telescopic rod 13 is provided. The telescopic rod 13 can be adjusted in length. During use, rotate the extension rod 133 to drive the adjusting screw rod 132 to rotate. In this way, the adjusting screw rod 132 will rotate out of the fixed rod 131, thereby extending the length of the telescopic rod 13 to meet more usage requirements and providing convenience for the user's work.

[0088] It should be noted that there are a large number of mature technologies to support the functions to be realized by each hardware in the present invention, which belong to the prior art. The essence of the present invention lies in optimizing the combination of the existing hardware and its connection mode for a specific application occasion to meet the adaptation requirements in the specific application occasion and solve the problems raised in the background art (without involving the improvement of the software inside the hardware).

[0089] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An AI recognition and early warning method that updates the surrounding environment in real time to improve recognition accuracy, characterized in that the recognition and capture are as follows:

1. Environmental change perception module:

1. Physical environment perception: Analyze changes in the surrounding environment (such as tree growth, grass cover, land morphology changes, etc.) through image recognition algorithms, and detect physical changes in the surrounding environment by comparing current images with historical images. For example, by calculating the relative position of trees and power lines, determine whether trees may affect the safety of power lines; 2. Sensor data collection: In addition to image data, the system can also combine environmental sensors (such as temperature, humidity, and air pressure sensors) and location sensors (such as GPS, lidar, etc.) to further analyze changes in the surrounding environment; 2. Data collection and preprocessing:

1. Image data acquisition: The camera regularly collects environmental images to capture dynamic environmental information such as tree growth and grass changes; 2. Environmental change recognition: Through image processing and comparative analysis, it can identify environmental changes in images, such as whether trees have grown taller or whether there are new obstacles (such as new buildings, changes in grass cover, etc.); 3. Incremental data processing: Incremental preprocessing and enhancement of image data in real time based on detected environmental changes to ensure that subsequent training input data contains features of the current environment; 3. Initial YOLO model deployment:

1. Deploy the YOLO model trained based on the existing environment (such as the initial state of trees and grass) on the terminal device. The model can identify common dangerous objects (such as branches, flying birds, etc.) and provide preliminary danger warnings; 4. Environmental change detection and incremental training mechanism:

1. Environmental change detection: By comparing with historical image data, the system analyzes changes in the surrounding environment in real time. If the environment changes significantly (such as trees becoming taller, grass coverage changing, etc.), the system will trigger an incremental training process; 2. Incremental training process: Through the incremental training algorithm, the YOLO model is fine-tuned based on the latest environmental data. The focus of incremental training is on partial updating of model weights rather than retraining from scratch to avoid wasting computing resources; 2.1 For objects related to environmental changes (such as tree occlusion, blocked wires, etc.), the system will make special adjustments to the model based on the actual environmental change data; 5. Adaptive model update and feedback mechanism:

1. Real-time feedback and adaptive adjustment: Based on the incremental training results caused by environmental changes, the system will evaluate the effect of model updates in real time. If the new model version can better identify the risks brought about by environmental changes (such as occlusions, obstacles, etc.), the system will automatically switch to the new version; 2. Model evaluation and rollback: When the system detects that the performance of the new model is not as expected (such as an increase in false positives and false negatives), it can quickly roll back to the previous stable version through the feedback mechanism to ensure the stability of the system; 6. Fusion of environmental data and image data:

1. Environmental data fusion: In addition to image data, the system will combine environmental data (such as humidity, wind speed, temperature, etc.) to analyze and judge the impact of environmental changes. For example, high humidity weather may cause trees and grass to grow faster. The system can combine this data to predict changes in advance and adjust the model accordingly.

2. Combined input of real-time images and environmental data: During incremental training, the fusion input of environmental data and image data will help the model more accurately perceive risk points related to environmental changes; 7. Distributed Update and Synchronization:

1. Distributed training and updating: When multiple terminal devices share the same area, the updated model will be synchronized to other devices through the cloud or edge computing network, ensuring that all devices use the same updated model when processing the same environmental changes.

2. Multi-terminal collaboration: When multiple terminal devices work together, they can share incrementally trained models and dynamically optimize based on environmental changes in different regions.

2. According to claim 1, an AI recognition and early warning method for real-time updating of surrounding environment to improve recognition accuracy is characterized by: In step six, dynamic environmental change perception: through image analysis and sensor data, real-time detection of physical changes in the surrounding environment (such as tree growth, changes in grass coverage, etc.), and adjustment of the model according to these changes to enhance the system's recognition capabilities in dynamic environments.

3. According to claim 1, the AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy is characterized by: In step six, incremental training and environmental adaptation: through the incremental training mechanism, not only new objects are trained, but also fine-tuning is performed according to environmental changes (such as changes in trees blocking power lines, changes in grass, etc.), thereby ensuring the adaptability of the model to the new environment.

4. According to claim 1, the AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy is characterized by: In step six, the real-time model update and feedback mechanism: triggers real-time update and model fine-tuning according to environmental changes, optimizes the recognition effect of the model in complex environments, and avoids false positives and false negatives caused by environmental changes.

5. According to claim 1, the AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy is characterized by: In step six, the environmental perception and automated feedback mechanism: automatically determine whether there are environmental changes through environmental data (such as image change analysis, sensor feedback, etc.), trigger incremental training and automatically adjust the model to ensure that the model always adapts to the latest environmental status.

6. The AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy according to claim 1, characterized in that: In the environmental change detection of step 4, 1., if it is found that trees are growing and blocking the transmission lines, the system will record the new growth of the trees and update the model to identify the new tree shape or relative position. If the grass coverage changes and affects the safety monitoring of the transmission lines, the system will also re-label and train the relevant data.

7. An AI recognition device for real-time updating of the surrounding environment to improve recognition accuracy, which is implemented by the AI ​​recognition warning method for real-time updating of the surrounding environment to improve recognition accuracy according to claims 1-6, comprising a bracket (4), characterized in that: A pan / tilt head (5) is fixedly mounted on the top of the bracket (4), a control terminal (7) is fixedly mounted on the top of the pan / tilt head, a fill light module (8) and a picture capture module (9) are fixedly mounted on both sides of the control terminal (7), a bracket (10) is fixedly mounted on the bottom of the bracket (4), and radars (11) are fixedly mounted on the bottoms of both sides of the bracket (10).

8. The AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy according to claim 7, characterized in that: A cross bar (3) is detachably mounted inside the bracket (4); an adaptor frame (2) is fixedly mounted on one end of the cross bar (3) away from the bracket (4); and a mounting frame (1) is fixedly mounted on the inner side of the adaptor frame (2).

9. The AI ​​recognition and early warning method for real-time updating of the surrounding environment to improve recognition accuracy according to claim 8, characterized in that: A plurality of mounting sleeves (12) are sleeved on the surface of the crossbar (3), and a plurality of telescopic rods (13) are movably mounted on the surface of one of the mounting sleeves (12).

10. The AI ​​recognition and early warning method for real-time updating of surrounding environment to improve recognition accuracy according to claim 9, characterized in that: The telescopic rod (13) comprises a fixed rod (131) movably mounted on the mounting sleeve (12); an adjusting screw rod (132) is threadedly mounted on one end of the fixed rod (131); and an end of the adjusting screw rod (132) away from the fixed rod (131) is fixedly connected to an extension rod (133).

Images