mahyarnajibi/SNIPER
A multi-scale deep learning object detector with efficient training (SNIPER) and inference (AutoFocus) algorithms published at NeurIPS 2018.
Not currently ranked — collecting fresh signals.
star history
SNIPER is an efficient multi-scale training approach for object detection that selectively processes context regions called chips around ground-truth objects, enabling larger batch sizes and GPU-efficient training with batch normalization. AutoFocus is an efficient multi-scale inference algorithm that adopts a coarse-to-fine approach, predicting category-agnostic segmentation maps to identify regions likely to contain small objects before processing them at finer scales.
Frequently asked
- What is mahyarnajibi/SNIPER?
- A multi-scale deep learning object detector with efficient training (SNIPER) and inference (AutoFocus) algorithms published at NeurIPS 2018.
- Is SNIPER open source?
- Yes — mahyarnajibi/SNIPER is an open-source project tracked on heatdrop.
- What language is SNIPER written in?
- mahyarnajibi/SNIPER is primarily written in Python.
- How popular is SNIPER?
- mahyarnajibi/SNIPER has 2.7k stars on GitHub.
- Where can I find SNIPER?
- mahyarnajibi/SNIPER is on GitHub at https://github.com/mahyarnajibi/SNIPER.