FoundationVision/VNext
A video instance recognition framework built on Detectron2 implementing state-of-the-art computer vision models for video segmentation and object tracking.
Not currently ranked — collecting fresh signals.
star history
VNext is a next-generation video instance recognition framework that provides advanced online and offline video instance segmentation algorithms along with motion models for object-centric video tasks. It officially implements multiple award-winning CVPR/ECCV papers including InstMove, SeqFormer, and IDOL, with IDOL winning first place in the video instance segmentation track of the 4th Large-scale Video Object Segmentation Challenge. The framework supports transformer-based architectures for video understanding tasks.
Frequently asked
- What is FoundationVision/VNext?
- A video instance recognition framework built on Detectron2 implementing state-of-the-art computer vision models for video segmentation and object tracking.
- Is VNext open source?
- Yes — FoundationVision/VNext is open source, released under the Apache-2.0 license.
- What language is VNext written in?
- FoundationVision/VNext is primarily written in Python.
- How popular is VNext?
- FoundationVision/VNext has 617 stars on GitHub.
- Where can I find VNext?
- FoundationVision/VNext is on GitHub at https://github.com/FoundationVision/VNext.