cython_bbox is widely used in object detection tasks. It was presumably first implemented in Faster-RCNN. Since then, almost all object detection projects use the source code directly.
In order to use it in standalone code snippets or small projects, Samson Wang made it a pypi module. The code was totally borrowed from Faster-RCNN. Thanks RBG and Samson!
In this repo, the code is extended to compute bbox overlaps together with an estimate of the corresponding uncertainties, provided that the spreads of the bboxes are given.
pip install git+https://github.com/DLR-MI/fuzzy_cython_bbox.git
from fuzzy_cython_bbox import fuzzy_bbox_ious
ious, ious_std = fuzzy_bbox_ious(
np.ascontiguousarray(atlbrs, dtype=float),
np.ascontiguousarray(btlbrs, dtype=float),
np.ascontiguousarray(axywh_stds, dtype=float),
np.ascontiguousarray(bxywh_stds, dtype=float)
)
For an example of how an ious matrix can be disambiguated run
python test/disambiguate.py