Spaces:
Running
Running
| title: PCB Defect Detector | |
| emoji: 🛠️ | |
| colorFrom: blue | |
| colorTo: green | |
| sdk: gradio | |
| sdk_version: 5.31.0 | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| # PCB Defect Detector | |
| A Hugging Face Space for **PCB defect detection and inspection reporting** using a **YOLOv8s** model fine-tuned on **DeepPCB**. | |
| ## What this app does | |
| Upload a PCB image to: | |
| - detect PCB defects with bounding boxes | |
| - identify the defect class | |
| - generate an inspection-style report with: | |
| - severity | |
| - explanation | |
| - likely root cause | |
| - possible impact | |
| - recommended action | |
| ## Supported defect classes | |
| - copper | |
| - mousebite | |
| - open | |
| - pin-hole | |
| - short | |
| - spur | |
| ## Model | |
| Base detector: **YOLOv8s** fine-tuned on DeepPCB | |
| Model card: [Janani-V/pcb-defect-yolov8s-deeppcb](https://huggingface.co/Janani-V/pcb-defect-yolov8s-deeppcb) | |
| ## Notes | |
| - The YOLO model performs **defect detection only**. | |
| - Severity, explanation, root cause, impact, and action are generated using a **rule-based defect knowledge layer** built on top of the model outputs. | |
| - This demo is intended for **educational, research, and prototype AOI workflows**. | |