Sebang EV Battery Inspection System with AI

U-Net segmentation · Weld measurement · Pinhole & spatter detection

Overview

EV battery welding inspection system using U-Net based segmentation for pinhole/spatter detection and weld size measurement. Critical for battery safety and quality in electric vehicle manufacturing.

🔋 EV Battery 🔬 U-Net 📏 Measurement Python PyTorch OpenCV

Problem

  • Battery welding defects critical for safety and performance.
  • Need to detect pinholes, spatter, and measure weld size.
  • Challenging imaging conditions with metallic surfaces.
  • Requires automated measurement for quality standards.

Solution

  • U-Net based segmentation for welding defect detection.
  • Automated measurement for weld size and defect severity.
  • Optimized for consistent industrial performance.
  • Multi-class defect recognition (pinholes, spatter, etc.).

My Role

  • Developed U-Net based segmentation for welding defect detection.
  • Implemented measurement for weld size and defect severity.
  • Optimized for consistent industrial performance.

Implementation

Language: Python

Libraries:

PyTorch U-Net OpenCV

Defects: Pinholes, spatter, weld size deviation

Results Gallery

EV battery inspection result

Demo Videos

Real-time inspection system in action.

End Result

  • Accurate welding defect detection with automated measurement output.
  • Improved EV battery quality and safety standards.
  • Deployed successfully in Sebang production line.