A newer version of the Gradio SDK is available:
6.1.0
metadata
title: X-YOLOv10
emoji: 🔍
colorFrom: indigo
colorTo: purple
sdk: gradio
sdk_version: 4.19.2
app_file: app.py
pinned: false
YOLOv10 Saliency Heat-map Visualiser
This Gradio app demonstrates object detection and saliency visualization using YOLOv10 models trained on the VOC dataset. The app allows users to:
- Choose between vanilla and finetuned YOLOv10 models
- Upload custom images or use provided examples
- Visualize object detections with bounding boxes
- See saliency heat-maps for each detected object
Models
- Vanilla VOC: Base YOLOv10 model trained on VOC dataset
- Finetune VOC: Fine-tuned YOLOv10 model with enhanced performance
Features
- Interactive web interface
- Real-time object detection
- Saliency heat-map generation
- Adjustable confidence threshold
- Example images included
Usage
- Select a model from the dropdown menu
- Upload an image or use one of the example images
- Adjust the confidence threshold if needed
- View the detection results and saliency heat-maps
Technical Details
The app uses:
- Gradio for the web interface
- YOLOv10 for object detection
- Custom feature extraction for saliency visualization
- OpenCV for image processing
Examples
The app includes two example images demonstrating the capabilities of the vanilla model.