The transmission line image defect diagnosis method based on a large model of power inspection uses a large model pre-training algorithm to design a complex, high-parameter network architecture and an industry big data training network to improve the model's perception capabilities and capture industry data characteristics more accurately. Improve the recognition accuracy of later application models. The self-supervised learning of massive data in the pre-training stage accumulates a large amount of background knowledge, allowing the model to infer richer semantic information from the image space context information when processing small samples, effectively improving the model's learning ability on small samples. and generalization ability. Design a foreground and background screening model, use 3D modeling with random backgrounds to produce inspection images, automatically segment and generate foreground and background image blocks, realize rapid training of the foreground and background screening model, realize pre-training data classification selection in real scenarios, and improve Large model pre-training speed. In the transfer learning process, a non-quantitative region-of-interest pooling layer is designed to achieve target positioning and classification optimization.
KEYWORDS: Education and training, Object detection, Visual process modeling, Transformers, Detection and tracking algorithms, Inspection, Data modeling
Aiming at the transmission defect business scenario with many components, various defect forms and uneven size distribution, this paper proposes a transmission defect identification algorithm based on the VIT pre-trained visual large model architecture and the ViTDet object detection training algorithm. Specifically, it uses the ViT-Large model as the backbone network and Cascade-rcnn as the framework of the ViTDet algorithm. Meanwhile, in order to solve the problem of small-size defect identification in transmission scenes with large field of view, the image clipping training strategy is integrated. Cut each image equally into four parts with an overlap rate of 20% during training. In the case of the similar false detection, the recognition rate of the large model has an improvement of about 5% compared to the traditional CNN model.
Due to the complex environmental background of the substation, the wide field of view of drone inspection, and the small proportion of the target in the image, it is difficult to extract the defect feature of the small target and the high false detection rate. How to improve the recognition rate of the defect of the small target equipment is an urgent problem to be solved at present. The multimodel cascaded module, which is used to locate device components before defect detection or defect classification, is introduced into the Cascade-Rcnn algorithm to improve the accuracy of defects of small target device components and achieve accurate location of device defects.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.