Arjun9211/steel-scratch-detector

Steel Scratch Detector ⚡

An industrial-grade tool for detecting and segmenting surface scratches on steel.

Overview

This application uses a high-precision ResUNet++ with Attention mechanism to analyze steel surface images and identify critical defects. It provides a real-time detection overlay and confidence metrics for each defect class.

Key Features

• —Instant Analysis: Drag and drop any steel surface image for immediate detection.
• —Deep Learning Core: Powered by PyTorch and ResUNet++ segmentation models.
• —Interactive Reports: Detailed confidence scores for each classified scratch.

How to use locally

1. 1.Install requirements: pip install -r requirements.txt
2. 2.Run the app: streamlit run streamlit_app.py