med333gg/fruit-detection-api

🍓 Fruit Detection API

A Hugging Face Space that runs YOLOv8 fruit detection on GPU for fast video processing.

Features

• —GPU Acceleration: Runs YOLOv8 on NVIDIA T4 GPU (10-50x faster than CPU)
• —Video Processing: Upload videos and get detection results
• —Real-time Results: See processed video with bounding boxes
• —API Access: Can be called from external applications

How to Use

1. 1.Web Interface: Upload a video file and click "Process Video"
2. 2.API Access: Call the space programmatically from your FastAPI backend

API Integration

Your FastAPI backend can call this space via HTTP requests to get GPU-accelerated detection results.

Model

• —Model: YOLOv8n (nano) - fastest public YOLO model
• —Classes: Detects COCO classes (currently using 'apple' class 53 as strawberry)
• —GPU: NVIDIA T4 with 16GB VRAM

Performance

• —Speed: 10-50x faster than CPU processing
• —Free Tier: 16 hours/month of GPU usage
• —Limitations: Video size and processing time limits apply