Artificial Intelligence
VisionQA — Computer Vision Quality Inspection
VisionQA is a production computer vision system deployed on manufacturing lines to detect surface defects, misalignments, and assembly errors in real time. Built on a custom-trained YOLOv8 model fine-tuned on client-specific defect imagery, it processes 60fps camera feeds and triggers alerts within 40ms. The system integrates with existing PLC and MES infrastructure.
Computer VisionDeep LearningEdge AI
2025
6 months
Confidential
Key Features
- Real-time Detection: 60fps inference with 40ms alert latency
- Custom Dataset: 12,000+ labelled defect images across 6 defect types
- 98.3% Precision: Exceeds human inspector accuracy on validation set
- Edge Deployment: ONNX + TensorRT for Raspberry Pi 5 and Jetson Nano
- Dashboard: Live feed viewer with bounding boxes and defect history
- PLC Integration: Direct signal output to halt production on critical defects
Technology Stack
Framework
YOLOv8 + FastAPI
Platform
Edge + Cloud
Language
Python
App Type
Industrial AI
Tools & Services
PyTorchOpenCVONNXTensorRTRaspberry Pi 5
Ready to Build Something Like This?
Let's collaborate and bring your vision to life with production-grade engineering.