CVMI-Lab/SyncOOD

(ECCV 2024) Can OOD Object Detectors Learn from Foundation Models?

26
/ 100
Experimental

This project helps computer vision engineers and researchers create synthetic images containing new, unusual objects with bounding box annotations. It takes existing image datasets and a foundation model, then automatically generates diverse scene-level images with novel objects. The output is a dataset of synthetic out-of-distribution (OOD) images that can be used to train robust object detection models.

No commits in the last 6 months.

Use this if you need to train an object detection model to recognize unexpected or novel objects, but lack real-world examples of these 'out-of-distribution' items.

Not ideal if your primary goal is standard in-distribution object detection where all possible object classes are well-represented in your training data.

computer-vision object-detection synthetic-data AI-safety model-robustness
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 4 / 25

How are scores calculated?

Stars

24

Forks

1

Language

Python

License

Apache-2.0

Last pushed

Dec 07, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/CVMI-Lab/SyncOOD"

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