MI-YOLO: more information based YOLO for insulator defect detection

Abstract

Insulators that connect high-voltage transmission lines may experience various faults due to long-term exposure to the natural environment, leading to safety degradation and reliability issues in power grids. Therefore, detecting defective insulators through daily maintenance and long-term overhaul is crucial. We propose an insulator defect detection method to achieve this objective, using a novel neural network named more information-you only look once (MI-YOLO). MI-YOLO includes several modifications compared to YOLOv5s: a skip connection module with downsampling in the backbone, a spatial pyramid dilated convolution module in the neck, and a novel serial-parallel spatial-channel attention module in between. These changes enhance the feature abstraction process by providing more information (MI) through the YOLO-like structure, hence the name MI-YOLO. To evaluate the superiority of MI-YOLO, multiple experiments are performed. First, an ablation study demonstrates the effectiveness of novel substructures and their combinations compared to YOLOv5s. Then, MI-YOLO is compared to baseline structures, including faster R-CNN, SSD, and five YOLOs. Furthermore, the performances of MI-YOLO and 10 state-of-the-art object detection methods are evaluated. Lastly, a second dataset is adopted to evaluate the generalization property and heat maps are presented. The experimental results show that the new designs substantially improve the accuracy of insulator fault detection by abstracting more information from the features.

Insulator Defect Detection Based on ML-YOLOv5 Algorithm

“…GC-YOLO integrates the Ghost convolution module into the backbone network, adds the CA attention mechanism, introduces the EVCBlock module in the neck layer, and includes an additional small object detection head in the detection layer. Luan et al [ 41 ] introduced MI-YOLO, an improved network based on YOLOv5, featuring a skip connection module with down-sampling in the backbone network, a neck layer with a spatial pyramid dilated convolution module, and the addition of a novel serial-parallel spatial-channel attention module. The evaluation metrics included precision (P), recall (R), mAP0.5, model parameters, FLOPs (Floating-Point Operations Per Second), and frames per second (FPS).…”

An Improved YOLOv8 Network for Detecting Electric Pylons Based on Optical Satellite Image

“…To obtain richer multi-scale feature representations and improve the detection accuracy and speed of the algorithm, the SPPF module [ 26 ] integrates information from multi-scale local features. This module provides additional information from a wider range of spatial levels [ 27 ], giving the network a global perspective. The SPPF module only needs to specify a convolutional kernel, and each pooling operation’s output serves as the input for the next, speeding up data processing [ 28 ].…”

WDC-YOLO: an improved YOLO model for small objects oriented printed circuit board defect detection

“…These changes provide more information and enhance the feature abstraction process. 15 Zhang et al integrate the inverse residual (IR) structure with the coordinate attention (CA) mechanism to construct a backbone network called CA Module for feature extraction. Additionally, they design a novel bidirectional weighted feature pyramid network (BWFPN) to improve the detection accuracy of the model.…”