Jamil Zakirov
Jamil Zakirov
Hi, thanks a lot for you job, idea described here is smart and simple! To apply it in my work I had to rewrite your code into class form. This...
**Problem** Currently most augmentation use simple `random.uniform(lower_bound, upper_bound)` sampling rule to select the parameters to applied. This makes all value in `[lower, upper]` range equally likely to occur. If someone,...
Proposed in paper _"Planckian jitter: enhancing the color quality of self-supervised visual representations"_ Link: https://arxiv.org/pdf/2202.07993.pdf [PyTorch code as a reference](https://kornia.readthedocs.io/en/latest/_modules/kornia/augmentation/_2d/intensity/planckian_jitter.html#RandomPlanckianJitter) Code from authors: https://github.com/TheZino/PlanckianJitter Example of ColorJitter vs PlanckianJitter from...
Hi, thanks again for the very useful library. I have a comment / feature requests to modify the behaviour of RGB shift augmentation. What it currently does is, basically creates...
Hello! Thanks for providing such an excellent library. While experimenting with it I miss one feature - ability to select final image shape for the whole batch. In PyTorch there...
Hi! When using `imagededup` with TF 2.5.0 I got the following _warning_ message: ``` /usr/local/lib/python3.8/dist-packages/tensorflow/python/keras/engine/training.py:2001: UserWarning: `Model.predict_generator` is deprecated and will be removed in a future version. Please use `Model.predict`,...
**Is your feature request related to a problem? Please describe.** deltaE measures distance between images in CIE Lab space. **Describe the solution you'd like** Re-write to PyTorch any of those...
**Is your feature request related to a problem? Please describe.** Very commonly used no-reference metric, which is mentioned in almost all IQA papers after it's initial release in 2013. **Describe...
Closes #261 ## Proposed Changes - [x] Move LPIPS and DISTS into separate files - [x] Mover test for LPIPS and DISTS into separate files - [x] Do not inherit...
**Describe the bug** When computing FSIM and FSIMc with device set to CUDA computation is still heavily CPU bounded. See screenshot of `htop` **To Reproduce** Steps to reproduce the behavior:...