Skip to content

Releases: stuartasiimwe7/DefectNet

Initial Release - v1.0

09 Oct 01:08
@stuartasiimwe7 stuartasiimwe7
v1.0
0a220db

Choose a tag to compare

View all tags
Initial Release - v1.0 Latest
Latest

AI-powered PCB defect detection system.

Features

  1. Real-time PCB defect detection using YOLOv5 deep learning models
  2. RESTful API built with FastAPI
  3. Rate limiting - 100 req/min for single predictions, 20 req/min for batch
  4. Batch processing - Process up to 10 images simultaneously
  5. Comprehensive testing - 32 passing tests with full coverage
  6. Docker containerization - Production-ready deployment

Detectable Defects

  • Missing components
  • Solder bridges
  • Open circuits
  • Short circuits
  • Component misalignment
  • Solder defects
  • Contamination

CI/CD Pipeline

  • Automated testing on every commit
  • Docker image building and publishing to GitHub Container Registry
  • Continuous integration with Python 3.12

Tech Stack

Built with FastAPI & Uvicorn for the REST API server, YOLOv5 for deep learning inference, Pillow & NumPy for image processing, pytest & pytest-mock for comprehensive testing, Docker for containerization, and SlowAPI for rate limiting.

Quick Start

Using Docker:

docker pull ghcr.io/stuartasiimwe7/defectnet:v1.0
docker run -p 8000:8000 ghcr.io/stuartasiimwe7/defectnet:v1.0

Local Setup:

git clone https://github.com/stuartasiimwe7/DefectNet
cd DefectNet
pip install -r requirements.txt
uvicorn app:app --host 0.0.0.0 --port 8000

Performance

  • Inference: ~45ms per image (CPU)
  • Throughput: ~20 images/second
  • Memory: ~500MB base

Documentation

Assets 2
Loading