Abstract:Routine visual inspection is fundamental to the preventive maintenance of power equipment. Convolutional neural networks (CNNs) substantially reduce the number of parameters and efficiently extract image features for classification tasks. In the actual production and operation process of substations, due to the limitation of safety distance, camera monitoring, inspection robots, etc., cannot be very close to the target. The operational environment of power equipment leads to scale variations in the main target and thus compromises the performance of conventional models. To address the challenges posed by scale fluctuations in power equipment image datasets, while adhering to the requirements for model efficiency and enhanced inter-channel communication, this paper proposed the ResNet Cross-Layer Parameter Sharing (ResNetCPS) framework. The core idea is that the network output should remain consistent for the same object at different scales. The proposed framework facilitates weight sharing across different layers within the convolutional network, establishing connections between pertinent channels across layers and leveraging the scale invariance inherent in image datasets. Additionally, for substation image processing mainly based on edge devices, smaller models must be used to reduce the expenditure of computing power. The Cross-Layer Parameter Sharing framework not only reduces the overall number of model parameters but also decreases training time. To further enhance the representation of critical features while suppressing less important or redundant ones, an Inserting and Adjacency Attention (IAA) module is designed. This mechanism improves the model’s overall performance by dynamically adjusting the importance of different channels. Experimental results demonstrate that the proposed method significantly enhances network efficiency, reduces the total parameter storage space, and improves training efficiency without sacrificing accuracy. Specifically, models incorporating the Cross-Layer Parameter Sharing module achieved a reduction in the number of parameters and model size by 10% to 30% compared to the baseline models.

Power Equipment Recognition Method Based on Mask R-CNN and Bayesian Context Network

Object recognition for power equipment via human-level concept learning

A Simultaneous Object Detection and Component Segmentation Approach Based on Mask R-CNN

Multi-label image recognition for electric power equipment inspection based on multi-scale dynamic graph convolution network

Dual-channel convolutional neural network for power edge image recognition

Research on power equipment recognition method based on image processing

High-resolution power equipment recognition based on improved self-attention

ID-Net: an improved mask R-CNN model for intrusion detection under power grid surveillance

A Mask RCNN based Automatic Reading Method for Pointer Meter

An 3D Point Cloud Reconstruction Approach from Single Image for Smart Power Plant

Resnetcps for Power Equipment and Defect Detection

Electrical Equipment Classification Via Improved Faster Region-based Convolutional Neural Network

Target Detection of Substation Electrical Equipment from Infrared Images Using an Improved Faster Regions with Convolutional Neural Network Features Algorithm

Data Recognition for Multi-Source Heterogeneous Experimental Detection in Cloud Edge Collaboratives

Intelligent Video Recognition of Violation Behavior in Substation Site Based on Mask R-CNN

Bimodal-based Object Detection and Instance Segmentation Models for Substation Equipments

Instance Segmentation Model for Substation Equipment Based on Mask R-CNN

Mask is All You Need: Rethinking Mask R-CNN for Dense and Arbitrary-Shaped Scene Text Detection

MFCP-Net: Multiscale Feature Correlation Perception Network for Image Segmentation Based on Background Prior

Image Denoising Technology of Power Equipment Based on Deep Residual Network

A remote sensing image target recognition method based on improved Mask-RCNN model