How josecerejo/solar-panel-defect-detector works
This repository is a small YOLOv8 solar-panel defect detection demo. A user submits an image and confidence threshold through Gradio; the app runs inference with the bundled weights, annotates detections, and returns an image and count. Dataset auditing, training, and evaluation are described in the README, but their implementation is not present in the supplied tree, so they are not modeled as runtime components.
Main entry points: app.py Demo app · best.pt Model weights
Components 7
Other components
- User
- Gradio UI
- Demo app
app.py - YOLO inference
- Model weights
best.pt - Annotated image
- Detection count
Connections 10
- User → Gradio UI: submits image
- Gradio UI → Demo app: calls predict
- Demo app → YOLO inference: runs inference
- Demo app → Model weights: loads weights
- YOLO inference → Demo app: returns results
- Demo app → Annotated image: plots detections
- Demo app → Detection count: counts boxes
- Annotated image → Gradio UI: displays image
- Detection count → Gradio UI: displays count
- Gradio UI → User: shows outputs