[KongchangAI]
Product

haralickt

haralickt is a PyTorch-based library for GPU-accelerated computation of Haralick texture features (GLCM). It supports all 13 descriptors, sliding windows, block tiling and batching.

Timeline (last 90 days)

Nov 13

haralickt is a pure PyTorch implementation library that computes all 13 Haralick descriptors on the GPU (also supporting multithreaded CPU), accelerated through vectorized tensor operations and tiled chunking.

Unverified50%
Nov 13

Compared with single-threaded mahotas, haralickt's CPU mode achieved an average speedup of 16.7x, ranging from 9.1x to 67.7x, in tests of 27 parameter combinations on 256x256 images with 32 gray levels.

Unverified50%
Nov 13

On image sizes of 512x512 and above, haralickt GPU's speedup relative to 10-thread OpenMP mahotas is roughly between 20 and 60 times.

Unverified50%
Nov 13

The maximum numerical difference between haralickt and mahotas results is 2.02 × 10⁻⁴, originating from float32 rounding error in Diff Variance.

Unverified50%
Nov 13

haralickt is a pure Python + PyTorch (>=2.0) implementation, and installation requires no C compiler toolchain.

Unverified50%
Nov 13

haralickt can be integrated with torchvision and MONAI, making it suitable for medical imaging deep learning pipelines.

Unverified50%
Nov 13

haralickt provides a sliding-window API `haralick(image, num_levels=32, window_size=32, stride=8, device="cuda")`, returning a tensor of shape (H, W, 13).

Unverified50%
Nov 13

haralickt provides `features.haralick()` as a drop-in replacement for mahotas, returning a NumPy array of shape (4, 13).

Unverified50%
Nov 13

haralickt's license is MPL-2.0.

Unverified50%
Nov 13

haralickt can be installed via `pip install haralickt`.

Unverified50%

2 more timeline events

All Facts (12)

Unverified

haralickt is a pure PyTorch implementation library that computes all 13 Haralick descriptors on the GPU (also supporting multithreaded CPU), accelerated through vectorized tensor operations and tiled chunking.

50%
Unverified

Compared with single-threaded mahotas, haralickt's CPU mode achieved an average speedup of 16.7x, ranging from 9.1x to 67.7x, in tests of 27 parameter combinations on 256x256 images with 32 gray levels.

50%
Unverified

On image sizes of 512x512 and above, haralickt GPU's speedup relative to 10-thread OpenMP mahotas is roughly between 20 and 60 times.

50%
Unverified

The maximum numerical difference between haralickt and mahotas results is 2.02 × 10⁻⁴, originating from float32 rounding error in Diff Variance.

50%
Unverified

haralickt is a pure Python + PyTorch (>=2.0) implementation, and installation requires no C compiler toolchain.

50%
Unverified

haralickt can be integrated with torchvision and MONAI, making it suitable for medical imaging deep learning pipelines.

50%
Unverified

haralickt provides a sliding-window API `haralick(image, num_levels=32, window_size=32, stride=8, device="cuda")`, returning a tensor of shape (H, W, 13).

50%
Unverified

haralickt provides `features.haralick()` as a drop-in replacement for mahotas, returning a NumPy array of shape (4, 13).

50%
Unverified

haralickt's license is MPL-2.0.

50%
Unverified

haralickt can be installed via `pip install haralickt`.

50%
Unverified

haralickt's current public benchmarks focus on 256x256 images and a limited set of parameter combinations; performance at larger sizes and more gray levels still needs community verification.

50%
Unverified

For processing small images, haralickt's speedup relative to CPU libraries is limited; large images and batch-processing scenarios are where the GPU truly pulls ahead.

50%

Source Articles