All Versions
19
Latest Version
Avg Release Cycle
56 days
Latest Release
69 days ago

Changelog History
Page 1

  • v0.5.6 Changes

    July 12, 2021

    โž• Added

    • โž• Added mix augmentations in containers (#1139)

    ๐Ÿ›  Fixed

    • ๐Ÿ›  Fixed non-4-dim input error for sequential (#1146)

    ๐Ÿ”„ Changed

    • Moving bbox-related functionality to bbox module (#1103)
    • Optimized version of hls_to_rgb and rgb_to_hls (#1154)

    โœ‚ Removed

    • โœ‚ Remove numpy dependency (#1136)
  • v0.5.5 Changes

    June 26, 2021

    โž• Added

    • โž• Added Stereo camera class (#1102)
    • โž• Added auto-generated images in docs (#1105) (#1108) (#1127) (#1128) (#1129) (#1131)
    • โž• Added chinese version README (#1112)
    • โž• Added random_apply to augmentaton containers (#1125)

    ๐Ÿ”„ Changed

    • ๐Ÿ”„ Change GaussianBlur to RandomGaussianBlur (#1118)
    • โšก๏ธ Update ci with pytorch 1.9.0 (#1120)
    • ๐Ÿ”„ Changed option for mean and std to be tuples in normalization (#987)
    • โœ… Adopt torch.testing.assert_close (#1031)

    โœ‚ Removed

    • โœ‚ Remove numpy import (#1116)
  • v0.5.4 Changes

    June 11, 2021

    โž• Added

    • โž• Add Canny edge detection (#1020)
    • โž• Added Batched forward function (#1058)
    • โž• Added denormalize homography function (#1061)
    • โž• Added more augmentations containers (#1014)
    • โž• Added calibration module and Undistort 2D points function (#1026)
    • โž• Added patch augmentation container (#1095)

    ๐Ÿ›  Fixed

    • โœ‚ Remove lena (#1059) :)

    ๐Ÿ”„ Changed

    • Resize regardless of number of dims, considering the last two dims as image (#1047)
    • ๐Ÿšฉ Raise error if converting to unit8 image to gray with float weights (#1057)
    • Filter 2D->2d, 3D->3d (#1069)
    • โœ‚ Removed augmentation functional module. (#1067)
    • ๐Ÿ‘‰ Make Morphology compatible with both OpenCV and Scipy (#1084)
  • v0.5.3 Changes

    May 29, 2021

    โž• Added

    • โž• Added inverse for augmentations (#1013)
    • โž• Add advanced augmentations: RandomFisheye, RandomElasticTransform, RandomThinPlateSpline, RandomBloxBlur (#1015

    ๐Ÿ›  Fixed

    • โœ… Correct Sobel test_noncontiguous. Nothing was tested before. (#1018)
    • Fixing #795: find_homography_dlt_iterated sometimes fails (#1022)

    ๐Ÿ”„ Changed

    • โ™ป๏ธ Refactorization of the morphology package (#1034)
    • Optimised clipping in clahe and some other minor optimisation (#1035)
  • v0.5.2

    May 14, 2021
  • v0.5.1 Changes

    April 30, 2021

    โž• Added

    • โž• Added dtype for create_mesh (#919)
    • โž• Added Hardnet8 (#955)
    • โž• Added normalize boolean for remap (#921)
    • โž• Added custom weights option for rgb2gray (#944)
    • โž• Added fp16 support (#963)
    • โž• Added ImageToTensor module and resize for non-batched images (#978)
    • โž• Add more augmentations (#960)
    • Anti alias resize (#989)
  • v0.4.1 Changes

    October 20, 2020

    ๐Ÿš€ Kornia 0.4.1 release

    Highlights

    We include new features for 3D augmentations:

    • RandomCrop3D
    • CenterCrop3D
    • RandomMotionBlur3D
    • RandomEqualize3D

    Few more core functionalities to work on 3D volumetric tensors:

    • warp_affine3d
    • warp_perspective3d
    • get_perspective_transform3d
    • crop_by_boxes3d
    • motion_blur3d
    • equalize3d
    • warp_grid3d

    Details changes

    โž• Added

    • Update docs for get_affine_matrix2d and get_affine_matrix3d (#618)
    • โž• Added docs for solarize, posterize, sharpness, equalize (#623)
    • โž• Added tensor device conversion for solarize params (#624)
    • โž• Added rescale functional and transformation (#631)
    • โž• Added Mixup data augmentation (#609)
    • โž• Added equalize3d (#639)
    • โž• Added decompose 3x4projection matrix (#650)
    • Added normalize_min_max functionality (#684)
    • โž• Added random equalize3d (#653)
    • โž• Added 3D motion blur (#713)
    • โž• Added 3D volumetric crop implementation (#689)
      • warp_affine3d
      • warp_perspective3d
      • get_perspective_transform3d
      • crop_by_boxes3d
      • warp_grid3d

    ๐Ÿ”„ Changed

    • Replace convolution with unfold in contrib.extract_tensor_patches (#626)
    • โšก๏ธ Updates Affine scale with non-isotropic values (#646)
    • Enabled param p for each augmentation (#664)
    • Enabled RandomResizedCrop batch mode when same_on_batch=False (#683)
    • Increase speed of transform_points (#687)
    • Improves find_homography_dlt performance improvement and weights params made optional (#690)
    • Enable variable side resizing in kornia.resize (#628)
    • โž• Added Affine transformation as nn.Module (#630)
    • Accelerate augmentations (#708)

    ๐Ÿ›  Fixed

    • Fixed error in normal_transform_pixel3d (#621)
    • ๐Ÿ›  Fixed pipelining multiple augmentations return wrong transformation matrix (#645)(645)
    • ๐Ÿ›  Fixed flipping returns wrong transformation matrices (#648)
    • ๐Ÿ›  Fixed 3d augmentations return wrong transformation matrix (#665)
    • ๐Ÿ›  Fix the SOSNet loading bug (#668)
    • ๐Ÿ›  Fix/random perspective returns wrong transformation matrix (#667)
    • ๐Ÿ›  Fixes Zca inverse transform (#695)
    • ๐Ÿ›  Fixes Affine scale bug (#714)

    โœ‚ Removed

    • โœ‚ Removed warp_projective (#689)

    Contributors

    @gaurav104 @shijianjian @mshalvagal @pmeier @ducha-aiki @qxcv @FGeri @vribeiro1 @ChetanPatil28 @alopezgit @jatentaki @dkoguciuk @ceroytres @ag14774

  • v0.4.0 Changes

    August 06, 2020

    ๐Ÿš€ Kornia 0.4.0 release

    ๐Ÿš€ In this release we are including the following main features:

    • ๐Ÿ‘Œ Support to PyTorch v1.6.0.
    • Local descriptors matching, homography and epipolar geometry API.
    • 3D augmentations and low level API to work with volumetric data.

    kornia_medical

    Highlights

    Local features matching

    โœ… We include an kornia.feature.matching API to perform local descriptors matching such classical and derived version of the nearest neighbour (NN).

    • โœ… match_nn
    • โœ… match_mnn
    • โœ… match_snn
    • โœ… match_smnn

      import torchimport kornia as Kdesc1 = torch.rand(2500, 128)desc2 = torch.rand(2500, 128)dists, idxs = K.feature.matching.match_nn(desc1, desc2) # 2500 / 2500x2

    Homography and epipolar geometry

    โœ… We also introduce kornia.geometry.homography including different functionalities to work with homographies and differentiable estimators based on the DLT formulation and the iteratively-reweighted least squares (IRWLS).

    import torchimport kornia as Kpts1 = torch.rand(1, 8, 2)pts2 = torch.rand(1, 8, 2)H = K.find\_homography\_dlt(pts1, pts2, weights=torch.rand(1, 8)) # 1x3x3
    

    โœ… In addition, we have ported some of the existing algorithms from opencv.sfm to PyTorch under kornia.geometry.epipolar that includes different functionalities to work with Fundamental , Essential or Projection matrices, and Triangulation methods useful for Structure from Motion problems.

    3D augmentations and volumetric

    ๐Ÿš€ We expand the kornia.augmentaion with a series of operators to perform 3D augmentations for volumetric data BxCxDxHxW. In this release, we include the following first set of geometric 3D augmentations methods:

    • RandomDepthicalFlip3D (along depth axis)
    • RandomVerticalFlip3D (along height axis)
    • RandomHorizontalFlip3D (along width axis)
    • RandomRotation3D
    • RandomAffine3D

    The API for 3D augmentation work same as with 2D image augmentations:

    import torchimport kornia as Kx = torch.eye(3).repeat(3, 1, 1)aug = K.augmentation.RandomVerticalFlip3D(p=1.0)print(aug(x))tensor([[[[[0., 0., 1.], [0., 1., 0.], [1., 0., 0.]],\<BLANKLINE\> [[0., 0., 1.], [0., 1., 0.], [1., 0., 0.]],\<BLANKLINE\> [[0., 0., 1.], [0., 1., 0.], [1., 0., 0.]]]]])
    

    โœ… Finally, we introduce also a low level API to perform 4D features transformations kornia.warp_projective and extending the filtering operators to support 3D kernels kornia.filter3D.

    More 2d operators

    We expand as well the list of the 2D image augmentations based on the paper AutoAugment: Learning Augmentation Policies from Data.

    • Solarize
    • Posterize
    • Sharpness
    • Equalize
    • RandomSolarize
    • RandomPosterize
    • RandomShaprness
    • RandomEqualize

    ๐Ÿ‘Œ Improvements

    • โž• add zca whitening (#458)
    • โž• add epipolar geometry package (#569)
    • Jit warp perspective (#574)
    • Autoaugment functions. (#571)
    • Dog and fix features (#591)
    • implement filter3D (#575)
    • Implement warp_projective (#587)
    • ๐Ÿ”‹ Feature matching and H/F/E estimation for SFM (#552)
    • 3D augmentations (#592)

    ๐Ÿ’ฅ Breaking changes

    • ๐Ÿ‘€ Create kornia.enhance submodule (#614) -> see details in here

    ๐Ÿ› Bugs/Fixes

    • ๐Ÿ›  fixed affine 2d shearing matrix translations (#612)
    • ๐Ÿ‘ป Now SIFTdesc throws and exception when the input parameters are incompatible (#598)
    • back to group conv backend for filter2d (#600)
    • โšก๏ธ updates sosnet git paths (#606)

    ๐Ÿ“„ Docs

    • โšก๏ธ Updated doc & example for augmentation (#583)
    • ๐Ÿ›  fix Tversky equation (#579)
    • โš  clean docs warnings (#604)
    • โž• add kornia.geometry.homography docs (#608)
    • create kornia.geometry.subpix (#610)

    Dev

    • ๐Ÿ‘Œ improve conftest fixtures and remove device, dtype imports (#568)
    • โœ… pin versions for pytest plugins and fix flake8 issues (#580)
    • made kornia versions explicit to pytorch version (#597)
  • v0.3.2 Changes

    August 06, 2020

    ๐Ÿš€ Kornia 0.3.2 release

    ๐Ÿš€ This release is just a checkpoint for the features in v0.4.0 with support to PyTorch 1.5.1.

    ๐Ÿš€ To see the new set of features check the release notes for Kornia 0.4.0.

  • v0.3.1 Changes

    May 10, 2020

    ๐Ÿš€ Kornia 0.3.1 release

    ๐Ÿš€ This release mainly introduces the following items:

    โž• Add support to Python 3.8

    ๐Ÿ”ฆ Exposes and fixes issues around align_corners.

    ๐Ÿ‘Œ Improve testing infrastructure adding parametrize for different devices and dtype and flake8/mypy support throw pytest by caching intermediate results. Test usage example:

    โœ… pytest -v --device cpu,cuda --dtype float16,float32,float64 --flake8 --mypy

    ๐Ÿ‘Œ Improvements

    • โšก๏ธ Update to python 3.8 (#550)
    • ๐Ÿ‘Œ Improve testing framework (#560)
    • ๐Ÿ›  Local feature fixes and nms improvements (#545)
    • Random motion blur improvments (#562)

    ๐Ÿ›  Fixes

    • ๐Ÿ”ฆ Expose align_corners everywhere, where interpolation occurs (#546)
    • โœ… Soft-argmax test fixes, renaming and enables jit (#553)

    ๐Ÿ› Bugs

    • ๐Ÿ›  Fix tests in TestSpatialSoftArgmax2d (#544)

    ๐Ÿ“„ Docs

    • โšก๏ธ Updated docstring for augmentation module (#554)