meder411 / Pytorch Emdloss
PyTorch 1.0 implementation of the approximate Earth Mover's Distance
Stars: ✭ 82
Labels
Projects that are alternatives of or similar to Pytorch Emdloss
Torch sampling
Efficient reservoir sampling implementation for PyTorch
Stars: ✭ 68 (-17.07%)
Mutual labels: cuda
Build Deep Learning Env With Tensorflow Python Opencv
Tutorial on how to build your own research envirorment for Deep Learning with OpenCV, Python, Tensorfow
Stars: ✭ 66 (-19.51%)
Mutual labels: cuda
Tsne Cuda
GPU Accelerated t-SNE for CUDA with Python bindings
Stars: ✭ 1,120 (+1265.85%)
Mutual labels: cuda
Parenchyma
An extensible HPC framework for CUDA, OpenCL and native CPU.
Stars: ✭ 71 (-13.41%)
Mutual labels: cuda
Mpn Cov
@ICCV2017: For exploiting second-order statistics, we propose Matrix Power Normalized Covariance pooling (MPN-COV) ConvNets, different from and outperforming those using global average pooling.
Stars: ✭ 63 (-23.17%)
Mutual labels: cuda
Cuda Design Patterns
Some CUDA design patterns and a bit of template magic for CUDA
Stars: ✭ 78 (-4.88%)
Mutual labels: cuda
Ggnn
GGNN: State of the Art Graph-based GPU Nearest Neighbor Search
Stars: ✭ 63 (-23.17%)
Mutual labels: cuda
Deepjointfilter
The source code of ECCV16 'Deep Joint Image Filtering'.
Stars: ✭ 68 (-17.07%)
Mutual labels: cuda
Modulated Deform Conv
deformable convolution 2D 3D DeformableConvolution DeformConv Modulated Pytorch CUDA
Stars: ✭ 81 (-1.22%)
Mutual labels: cuda
2016 super resolution
ICCV2015 Image Super-Resolution Using Deep Convolutional Networks
Stars: ✭ 78 (-4.88%)
Mutual labels: cuda
Titan
A high-performance CUDA-based physics simulation sandbox for soft robotics and reinforcement learning.
Stars: ✭ 73 (-10.98%)
Mutual labels: cuda
PyTorch EMDLoss
PyTorch 1.0 implementation of the approximate Earth Mover's Distance
This is a PyTorch wrapper of CUDA code for computing an approximation to the Earth Mover's Distance loss.
Original source code can be found here. This repository updates the code to be compatible with PyTorch 1.0 and extends the implementation to handle arbitrary dimensions of data.
Installation should be as simple as running python setup.py install
.
Limitations and Known Bugs:
- Double tensors must have <=11 dimensions while float tensors must have <=23 dimensions. This is due to the use of CUDA shared memory in the computation. This shared memory is limited by the hardware to 48kB.
- When handling larger point sets (M, N > ~2000), the CUDA kernel will fail. I think this is due to an overflow error in computing the approximate matching kernel. Any suggestions to fix this would be greatly appreciated. I have pinpointed the source of the bug here.
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].