ghayas1214/Traffic-monitoring-app

YOLOv5 Video Object Detection

This application provides an easy-to-use interface for object detection in videos using YOLOv5.

Features

• —🎯 Detects 80+ object classes (people, vehicles, animals, etc.)
• —📹 Supports multiple video formats (MP4, AVI, MOV)
• —⚙️ Adjustable confidence threshold
• —📊 Real-time progress tracking
• —💾 Download processed videos with bounding boxes

How to Use

1. 1.Upload your video file
2. 2.Adjust the confidence threshold (default: 0.25)
3. 3.Lower = more detections (may include false positives)
4. 4.Higher = fewer but more accurate detections
5. 5.Click 'Process Video'
6. 6.Wait for processing to complete
7. 7.Download your processed video

Detected Object Classes

The YOLOv5 model can detect 80 COCO dataset classes including:

People & Animals:

• —Person, Dog, Cat, Horse, Bird, Sheep, Cow, Elephant, Bear, Zebra, Giraffe

Vehicles:

• —Car, Truck, Bus, Motorcycle, Bicycle, Train, Airplane, Boat

Everyday Objects:

• —Laptop, Cell Phone, Book, Clock, Bottle, Cup, Chair, Couch, Bed

Sports Equipment:

• —Baseball Bat, Baseball Glove, Skateboard, Surfboard, Tennis Racket

And many more!

Performance Tips

• —✅ For faster processing, use shorter videos (under 1 minute)
• —✅ Higher resolution videos take longer to process
• —✅ Increase confidence threshold for fewer detections
• —✅ The app works best with clear, well-lit videos

Technical Details

• —Model: YOLOv5s (Small) - Good balance between speed and accuracy
• —Framework: PyTorch + Ultralytics YOLOv5
• —Interface: Gradio
• —Processing: CPU-based (suitable for Hugging Face Spaces free tier)

Limitations

• —Processing time depends on video length and resolution
• —Large videos (>5 minutes) may timeout on free tier
• —CPU processing is slower than GPU but more accessible

Credits

• —YOLOv5: Ultralytics
• —Gradio: Hugging Face
• —Dataset: COCO Dataset

License

GPL-3.0 (inherited from YOLOv5)

Built with ❤️ using YOLOv5 and Gradio