dronefreak/VisDrone-dataset-python-toolkit

Modern PyTorch toolkit for the VisDrone aerial object detection dataset with production-ready training pipelines, real-time inference, and format converters. Features state-of-the-art models (Faster R-CNN, FCOS, RetinaNet), mixed precision training, rich progress tracking, and optimizations for small object detection in drone imagery.

56
/ 100
Established

This toolkit helps you analyze drone footage by automatically finding and labeling objects like cars, people, and bicycles in aerial images. You provide drone imagery (photos or video) and it outputs analyzed images with detected objects highlighted. This is ideal for anyone working with drone-collected data for surveillance, traffic monitoring, or environmental surveys who needs to identify specific items within that footage.

Use this if you need to accurately detect small objects like vehicles and pedestrians in drone-captured images and want a ready-to-use system to train and deploy detection models.

Not ideal if your primary use case involves detecting objects in standard ground-level photography or if you need to detect highly occluded objects or objects smaller than 15 pixels in aerial views.

drone-imagery aerial-surveillance traffic-monitoring object-detection remote-sensing
No Package No Dependents
Maintenance 13 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 19 / 25

How are scores calculated?

Stars

61

Forks

22

Language

Python

License

Apache-2.0

Last pushed

Mar 19, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/dronefreak/VisDrone-dataset-python-toolkit"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.