JosephKJ/OWOD
Open World Object Detector using contrastive clustering and energy-based unknown identification for incremental object detection.
Not currently ranked — collecting fresh signals.
star history
ORE is a CVPR 2021 Oral paper introducing the Open World Object Detection problem where a model must identify unknown objects without explicit supervision and incrementally learn new classes without forgetting. The solution uses contrastive clustering to separate known and unknown instances and energy-based scores for unknown identification. It is built on Detectron2 and achieves state-of-the-art results in incremental object detection settings.
Frequently asked
- What is JosephKJ/OWOD?
- Open World Object Detector using contrastive clustering and energy-based unknown identification for incremental object detection.
- Is OWOD open source?
- Yes — JosephKJ/OWOD is open source, released under the Apache-2.0 license.
- What language is OWOD written in?
- JosephKJ/OWOD is primarily written in Python.
- How popular is OWOD?
- JosephKJ/OWOD has 1.1k stars on GitHub.
- Where can I find OWOD?
- JosephKJ/OWOD is on GitHub at https://github.com/JosephKJ/OWOD.