Analysis on DeepLabV3+ Performance for Automatic Steel Defects Detection

2020/04/09 by Zheng Nie, Nie, Zheng, Jiachen Xu +3 · 6 citations
Computer Science · Engineering · Psychology · #Advanced Neural Network Applications #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Currency Recognition and Detection #Engineering #FOS: Computer and information sciences #Forensic engineering #Industrial Vision Systems and Defect Detection #Psychology #cs.CV

paper · pdf · doi:10.48550/arxiv.2004.04822

published in arXiv (Cornell University) (Cornell University) · 10 pages, 30 figures

openalex publication_date 2020/04/09 · arxiv created 2020/04/15 · arxiv updated 2020/04/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Our works experimented DeepLabV3+ with different backbones on a large volume of steel images aiming to automatically detect different types of steel defects. Our methods applied random weighted augmentation to balance different defects types in the training set. And then applied DeeplabV3+ model three different backbones, ResNet, DenseNet and EfficientNet, on segmenting defection regions on the steel images. Based on experiments, we found that applying ResNet101 or EfficientNet as backbones could reach the best IoU scores on the test set, which is around 0.57, comparing with 0.325 for using DenseNet. Also, DeepLabV3+ model with ResNet101 as backbone has the fewest training time.

Citations

• Densely Connected Convolutional Networks
• Fully Convolutional Networks for Semantic Segmentation

Related

• A Steel Surface Defect Detection Method Based on Lightweight Convolution Optimization
• One-class Steel Detector Using Patch GAN Discriminator for Visualising Anomalous Feature Map
• Comparative Analysis of Object Detection Algorithms for Surface Defect Detection
• YOLO-Based Defect Detection for Metal Sheets
• Efficient Neural Net Approaches in Metal Casting Defect Detection