An AI-based method for filtering image data in power scenarios

Abstract

This invention relates to the field of image filtering technology and discloses an artificial intelligence-based method for filtering power scene image data, comprising the following steps: Step 1: Marking images according to the device; Step 2: Marking images according to the shooting time; Step 3: Marking images according to the target subject and aggregating the above information; Step 4: Removing duplicate image data; Step 5: Artificial intelligence identifies and outputs the filtered images. This invention, before performing artificial intelligence filtering, first groups and marks all images according to different devices, and simultaneously marks image data based on different shooting times. Then, after identifying the target subject in the image, the above-mentioned marking content is aggregated into a QR code and displayed on the image. Finally, the artificial intelligence system can remove duplicate images of the same target taken by the same device before performing the filtering operation, avoiding the removal of non-duplicate image data and affecting the final filtering accuracy.

Description

Technical Field

[0001] This invention relates to the field of image filtering technology, specifically to an artificial intelligence-based method for filtering power scene image data. Background Technology

[0002] With the rapid development of big data and the intelligent Internet, the traditional power industry is also gradually realizing intelligentization. In accordance with the State Grid Corporation's overall strategic deployment for building a smart grid, the power industry is about to undergo revolutionary changes. With the development of the intelligent era, the power industry is continuously generating a large amount of data with rich data types. With the development of intelligent inspection, the ability to quickly and automatically identify images of power equipment defects from a large number of power inspection images collected by aerial photography or robots plays a significant role in timely fault repair.

[0003] Artificial intelligence (AI) systems enable machines to perform complex tasks that typically require human intelligence. They can replace manual labor in filtering image data of power scenarios, removing images of equipment operating normally, retaining images of faulty equipment, and then having them verified manually. This can significantly reduce workload and quickly identify equipment faults.

[0004] The implementation of this technical solution has at least the following drawbacks: Power inspection equipment captures a large number of images during operation. While the AI ​​system removes duplicate images to improve efficiency, similar images of different equipment or cables, or images of the same equipment captured at different times, may be mistakenly identified as duplicates or similar and deleted. This could lead to the exclusion of important images during screening, affecting the accuracy of the screening process. Therefore, we propose an AI-based method for filtering power scene image data. Summary of the Invention

[0005] The purpose of this invention is to provide an artificial intelligence-based method for filtering power scene image data. Before performing AI filtering, all images are grouped and marked according to different devices. At the same time, the image data is marked according to the different shooting times. After identifying the target subject in the image, the above-mentioned marked content is combined into a QR code and displayed on the image. Finally, the AI ​​system can eliminate duplicate images of the same target taken by the same device before performing the filtering operation. This avoids the elimination of non-duplicate image data, which would affect the final filtering accuracy, and solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for filtering power scene image data based on artificial intelligence, comprising the following steps:

[0007] Step 1: First, group all the image data according to the different inspection equipment, and mark the image data in the same group with the same equipment label information, and record the information at the same time;

[0008] Step 2: In the same group of image data, mark the corresponding shooting time on the corresponding image data according to the different shooting times of different image data, and also record the specific shooting time;

[0009] Step 3: Use an artificial intelligence system to perform preliminary identification on all image data, determine the main target on the image data and identify the marker, and combine the marker of the main target with the above-mentioned device marker information and time marker information, and display it in the form of a QR code at the corner of the image data;

[0010] Step 4: Use the artificial intelligence system to identify the image data again. Quickly read the data such as the shooting device, time and subject of the image through QR code. When there are two or more pictures with the same shooting device, short time interval and the same subject, only one picture is kept as a sample. Otherwise, the image data is not removed, thereby reducing the total number of samples.

[0011] Step 5: Finally, use an artificial intelligence system to compare images of different target subjects with normal working images of the corresponding target subjects, filter out target images that may have faults, and output them.

[0012] In a preferred embodiment of the present invention, the total number of groups in step one is the same as the number of inspection devices, and the inspection devices can be one or more imaging devices.

[0013] In a preferred embodiment of the present invention, the inspection equipment in step one can be one or more of the following: drone, inspection robot, track robot, etc.

[0014] In a preferred embodiment of the present invention, the output image in step five is the final filtered image data, which can be manually checked to determine whether there are any faults.

[0015] In a preferred embodiment of the present invention, the artificial intelligence systems in steps three to four are all trained with deep learning and can achieve automatic target recognition.

[0016] In a preferred embodiment of the present invention, the normal working image in step five is obtained by daily sampling and screening of qualified images by the inspection equipment, and the screened images are stored in the database for use at any time.

[0017] In a preferred embodiment of the present invention, all processes in steps one to five are performed in a computer containing an artificial intelligence system.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0019] This invention first groups and marks all images according to different devices before performing artificial intelligence screening. At the same time, it marks the image data according to the different shooting times. After identifying the target subject of the image, the above-mentioned marking content is combined into a QR code and displayed on the image. Finally, the artificial intelligence system can eliminate duplicate images of the same target taken by the same device before performing the screening operation, so as to avoid the elimination of non-duplicate image data and affect the final screening accuracy. Attached Figure Description

[0020] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0021] Figure 1 This is a flowchart of an artificial intelligence-based method for filtering power scene image data according to the present invention. Detailed Implementation

[0022] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0023] Example 1

[0024] Please see Figure 1 This invention provides a technical solution: a method for filtering power scene image data based on artificial intelligence, comprising the following steps:

[0025] Step 1: Use inspection equipment to capture images of electrical equipment. First, group all the image data according to the different inspection equipment. The total number of groups is the same as the number of inspection equipment. The inspection equipment can be one or more imaging devices, such as drones, inspection robots, track robots, etc. Mark the image data of the same group with the same equipment label information and record the information.

[0026] Step 2: In the same group of image data, mark the corresponding shooting time on the corresponding image data according to the different shooting times of different image data, and also record the specific shooting time;

[0027] Step 3: Use an artificial intelligence system to perform preliminary identification on all image data, determine the main target on the image data and identify the marker, and combine the marker of the main target with the above-mentioned device marker information and time marker information, and display it in the form of a QR code at the corner of the image data;

[0028] Step 4: Use the artificial intelligence system to identify the image data again. Quickly read the data such as the shooting device, time and subject of the image through QR code. When there are two or more pictures with the same shooting device, short time interval and the same subject, only one picture is kept as a sample. Otherwise, the image data is not removed, thereby reducing the total number of samples.

[0029] Step 5: Finally, the artificial intelligence system is used to compare the images of different target subjects with the normal working images of the corresponding target subjects, filter out the target images that may have faults and output them. The output images are the final filtered image data. After manual inspection, it can be determined whether there are fault problems, and the fault problems of electrical equipment can be judged relatively accurately.

[0030] The artificial intelligence systems in steps three and four have all undergone deep learning training and can achieve automatic target recognition. During the learning process, the artificial intelligence system gradually improves its recognition accuracy through the input of a large amount of image data.

[0031] The normal working images in step five are obtained by screening qualified images through daily sampling by the inspection equipment, and the screened images are stored in the database for use at any time. The image database is continuously updated as the artificial intelligence system learns deeply.

[0032] All processes in steps one through five are completed in a computer containing an artificial intelligence system.

[0033] Example 2

[0034] Please see Figure 1 This invention provides a technical solution: a method for filtering power scene image data based on artificial intelligence, comprising the following steps:

[0035] Step 1: Use inspection equipment to capture images of electrical equipment. First, group all the image data according to the different inspection equipment. The total number of groups is the same as the number of inspection equipment. The inspection equipment can be one or more imaging devices, such as drones, inspection robots, track robots, etc. Mark the image data of the same group with the same equipment label information and record the information for future use.

[0036] Step 2: Use an artificial intelligence system to perform preliminary identification on all image data, determine the main target in the image data and identify the marker, and record this information as well;

[0037] Step 3: In all the image data, according to the different shooting times of different image data, mark the corresponding shooting time on the corresponding image data, and also record the specific shooting time. Combine the time mark with the above-mentioned device marking information and subject target marking information, and display it in the corner of the image data in the form of a QR code, so that the artificial intelligence system can recognize and read this data.

[0038] Step 4: Use the artificial intelligence system to identify the image data again. Quickly read the data such as the shooting device, time and subject of the image through QR code. When there are two or more pictures with the same shooting device, short time interval and the same subject, only one of them is kept as a sample. Otherwise, if any of the above conditions are not met, the image data will not be removed, thereby reducing the total number of samples.

[0039] Step 5: Finally, the artificial intelligence system is used to compare the images of different target subjects with the normal working images of the corresponding target subjects, filter out the target images that may have faults and output them. The output images are the final filtered image data. After manual inspection, it can be determined whether there are fault problems, and the fault problems of electrical equipment can be judged relatively accurately.

[0040] The artificial intelligence systems in steps three and four have all undergone deep learning training and can achieve automatic target recognition. During the learning process, the artificial intelligence system gradually improves its recognition accuracy through the input of a large amount of image data.

[0041] The normal working images in step five are obtained by screening qualified images through daily sampling by the inspection equipment, and the screened images are stored in the database for use at any time. The image database is continuously updated as the artificial intelligence system learns deeply.

[0042] All processes in steps one through five are completed in a computer containing an artificial intelligence system.

[0043] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the scope of the invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0044] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for filtering power scene image data based on artificial intelligence, characterized in that, Includes the following steps: Step 1: First, group all the image data according to the different inspection equipment, and mark the image data in the same group with the same equipment label information, and record the information at the same time; Step 2: In the same group of image data, mark the corresponding shooting time on the corresponding image data according to the different shooting times of different image data, and also record the specific shooting time; Step 3: Use an artificial intelligence system to perform preliminary identification on all image data, determine the main target on the image data and identify the marker, and combine the marker of the main target with the above-mentioned device marker information and time marker information, and display it in the form of a QR code at the corner of the image data; Step 4: Use the artificial intelligence system to identify the image data again. Quickly read the image's shooting device, time, and subject data through QR code. When there are two or more images with the same shooting device, short time interval, and the same subject, only one image is kept as a sample. Otherwise, the image data is not removed, thus reducing the total number of samples. Step 5: Finally, use an artificial intelligence system to compare images of different target subjects with normal working images of the corresponding target subjects, filter out target images that may have faults, and output them.

2. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: The total number of groups in step one is the same as the number of inspection devices, which can be one or more imaging devices.

3. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: The inspection equipment in step one can be one or more of the following: drone, inspection robot, and track robot.

4. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: The output image in step five is the final filtered image data. After manual inspection, it can be determined whether there are any faults.

5. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: The artificial intelligence systems in steps three and four have all been trained with deep learning and can achieve automatic target recognition.

6. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: The normal working images in step five are obtained by daily sampling and screening of qualified images by the inspection equipment, and the screened images are stored in the database for use at any time.

7. The method for filtering power scene image data based on artificial intelligence according to claim 1, characterized in that: All processes in steps one through five are completed in a computer containing an artificial intelligence system.

Images