Ak130204/Object-Drone-Detection

Drone Detection Using YOLOv8n (COCO Pretrained + Custom Training) This project demonstrates real-time drone detection using YOLOv8n starting from a COCO-pretrained model (yolov8n.pt) and then training on a custom drone dataset to produce the final trained model best.pt.

Features Real-time drone detection using YOLOv8n

Starts from COCO-pretrained yolov8n.pt

Fine-tuned on a custom drone dataset → output model: best.pt

Supports image, video, and webcam inference

Exports models to ONNX, TensorRT, CoreML, etc.

Training YOLOv8n on Custom Drone Dataset Using COCO Pretrained YOLOv8n.pt

yolo detect train \model=yolov8n.pt \data=data.yaml \epochs=50 \imgsz=640

Output Model:

runs/detect/train/weights/best.pt

Inference (Using best.pt) Predict on an image: yolo detect predict model=runs/detect/train/weights/best.pt source="test.jpg"

Predict on a video: yolo detect predict model=runs/detect/train/weights/best.pt source="drone_video.mp4"

Real-time webcam detection: yolo detect predict model=runs/detect/train/weights/best.pt source=0