xingyizhou/CenterNet2
A two-stage object detector using class-agnostic one-stage detection as proposal network with probabilistic framework.
Not currently ranked — collecting fresh signals.
star history
CenterNet2 is an object detection architecture that combines one-stage and two-stage detection paradigms. The first stage estimates object probabilities using a class-agnostic detector as a proposal network, while the second stage conditionally classifies detected objects. Built on detectron2, it achieves state-of-the-art accuracy (56.4 mAP on COCO) while maintaining competitive inference speeds.
Frequently asked
- What is xingyizhou/CenterNet2?
- A two-stage object detector using class-agnostic one-stage detection as proposal network with probabilistic framework.
- Is CenterNet2 open source?
- Yes — xingyizhou/CenterNet2 is open source, released under the Apache-2.0 license.
- What language is CenterNet2 written in?
- xingyizhou/CenterNet2 is primarily written in Python.
- How popular is CenterNet2?
- xingyizhou/CenterNet2 has 1.2k stars on GitHub.
- Where can I find CenterNet2?
- xingyizhou/CenterNet2 is on GitHub at https://github.com/xingyizhou/CenterNet2.