This is our PyTorch implementation of the paper "LiteYOLO-ID: A Lightweight Object Detection Network for Insulator Defect Detection
" published in IEEE Transactions on Instrumentation and Measurement.
Install
First, clone the project and configure the environment. Python>=3.7.0, PyTorch>=1.7.
git clone https://github.com/LuYang-2023/Insulator-defect-detection.git # clone
cd Insulator-defect-detection
pip install -r requirements.txt # install
Train
python train.py --cfg models/LiteYOLO-ID.yaml --data data/mydata.yaml
Test
python val.py --data data/mydata.yaml --weights best.pt --task test
The lightweight convolutional module EGC incorporates the design philosophies of GhostNet and C2f modules, significantly enhancing the capture of key information in detection targets through the ECA attention mechanism. The structural diagram of the EGC module is shown below.
My dataset contains sensitive information from the Tianjin Power Grid Company, and I need to communicate with them first to confirm whether this data can be made public.
The hardware and software configuration of the Jetson TX2 NX includes an NVIDIA Pascal GPU, with PyTorch version 1.8.0 and CUDA version 10.2.
If you use this code or article in your research, please cite it using the following BibTeX entry:
@ARTICLE{10569022,
author = {Li, Dahua and Lu, Yang and Gao, Qiang and Li, Xuan and Yu, Xiao and Song, Yu},
title = {LiteYOLO-ID: A Lightweight Object Detection Network for Insulator Defect Detection},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2024},
volume = {73},
pages = {1-12},
doi = {10.1109/TIM.2024.3418082},
keywords = {Insulators, Accuracy, Computational modeling, Defect detection, YOLO, Feature extraction, Neck, Deep learning, insulator defect detection, lightweight, quantification, deployment}
}
Email:[email protected]