shinya7y/UniverseNet
UniverseNet provides state-of-the-art object detection models and a universal-scale benchmark (USB) for evaluating detection across varying object scales and image domains.
Not currently ranked — collecting fresh signals.
star history
UniverseNet is the official implementation of the USB benchmark published at BMVC 2022. It extends the MMDetection framework to provide state-of-the-art detectors called UniverseNets for universal-scale object detection. The benchmark incorporates COCO, Waymo Open Dataset, and Manga109-s to evaluate detection performance across varying object scales and image domains, with fair training and evaluation protocols.
Frequently asked
- What is shinya7y/UniverseNet?
- UniverseNet provides state-of-the-art object detection models and a universal-scale benchmark (USB) for evaluating detection across varying object scales and image domains.
- Is UniverseNet open source?
- Yes — shinya7y/UniverseNet is open source, released under the Apache-2.0 license.
- What language is UniverseNet written in?
- shinya7y/UniverseNet is primarily written in Python.
- How popular is UniverseNet?
- shinya7y/UniverseNet has 433 stars on GitHub.
- Where can I find UniverseNet?
- shinya7y/UniverseNet is on GitHub at https://github.com/shinya7y/UniverseNet.