Edgar Riba
Edgar Riba
## ๐ Feature We would like to have in `kornia.enhance` a batched device agnostic version of the `cv::threshold` including the different variants. Different threshold types: - [ ] THRESH_BINARY -...
#### Changes Fixes # (issue) #### Type of change - [ ] ๐ Documentation Update - [ ] ๐งช Tests Cases - [ ] ๐ Bug fix (non-breaking change which...
#### Changes Fixes #1747 #### Type of change - [ ] ๐ Documentation Update - [ ] ๐งช Tests Cases - [ ] ๐ Bug fix (non-breaking change which fixes...
squared_distance is quite innecifient used in this way. The funciton itself first compute the norm, then it is squared and now sqrt.... use directly the `norm` here. _Originally posted by...
Update the `FaceDetection` API based on the changes from `libfacedetection.train` API --> https://github.com/ShiqiYu/libfacedetection.train/commit/e5c4a24de8446bd1069b923a6f318d6602677ebd _Originally posted by @edgarriba in https://github.com/kornia/kornia/pull/1780#discussion_r914611891_
@shijianjian the `Boxes` validation test must be updated _Originally posted by @edgarriba in https://github.com/kornia/kornia/pull/1713#discussion_r913659580_
## ๐ Feature Implement the pytorch version of `cv::integral` https://docs.opencv.org/4.6.0/d7/d1b/group__imgproc__misc.html#ga97b87bec26908237e8ba0f6e96d23e28 Some reference: - https://github.com/scikit-image/scikit-image/blob/main/skimage/transform/integral.py - https://discuss.pytorch.org/t/is-there-an-elegant-way-to-calculate-integral-image-using-pytorch-api/22669/5 ______________________________________________________________________ #### Consider also to contribute to Kornia universe projects :) - [**Tutorials**](https://github.com/kornia/tutorials): our...
## ๐ Feature Deprecate python 3.6 -- support Python >= 3.7 ______________________________________________________________________ #### Consider also to contribute to Kornia universe projects :) - [**Tutorials**](https://github.com/kornia/tutorials): our repository containing the tutorials.
### Describe the bug Only happens with `torch.float32` / `torch.device("cuda")` For `torch.float64` / `torch.device("cuda")` seems fine ```bash > dst_norm_trans_src_norm: torch.Tensor = normalize_homography(M, (H, W), (h_out, w_out)) # Bx3x3 (Pdb) normalize_homography(M,...
## ๐ Feature Add `Quaternion` class that is compatible with `Eigen::Quaternion` https://eigen.tuxfamily.org/dox/classEigen_1_1Quaternion.html Initial implementation here: https://github.com/kornia/kornia/blob/liegroup-algebra/kornia/geometry/liegroup/quaternion.py ## Motivation We want a simple api that can be easily combined with `kornia.geometry.conversions`...