You can find full list of my open-source projects under my GitHub account: https://github.com/BloodAxe.
However, there are some projects I worked on that are worth mentioning here:
Pytorch-toolbelt is a Python library with a set of bells and whistles for PyTorch for fast R&D prototyping and Kaggle farming.
- Easy model building using flexible encoder-decoder architecture.
- Modules: CoordConv, SCSE, Hypercolumn, Depthwise separable convolution and more.
- GPU-friendly test-time augmentation TTA for segmentation and classification
- GPU-friendly inference on huge (5000x5000) images
- Every-day common routines (fix/restore random seed, filesystem utils, metrics)
- Losses: BinaryFocalLoss, Focal, ReducedFocal, Lovasz, Jaccard and Dice losses, Wing Loss and more.
- Extras for Catalyst library (Visualization of batch predictions, additional metrics)
- Showcase: Catalyst, Albumentations, Pytorch Toolbelt example: Semantic Segmentation @ CamVid
Why Honest answer is “I needed a convenient way to re-use code for my Kaggle career”. During 2018 I achieved a Kaggle Master badge and this been a long path. Very often I found myself re-using most of the old pipelines over and over again. At some point it crystallized into this repository.
This lib is not meant to replace catalyst / ignite / fast.ai high-level frameworks. Instead it’s designed to complement them.
This project is a fork of Albumentations. I was a member of Albumentations core team back in a day. Since the beginning of russian invasion into Ukraine I’ve left the team. You can find answers why here.
This fork contains few improvements/additions that I feel are necessary to have. Since this private fork is mainly meant for personal use I feel less obliged to keep backward compatibility with the original project.
At some point it will become deprecated and transform into something new & shiny.
Yet another fork of high-level training library around PyTorch. Forked quite a while ago to ensure API will remain stable and also added missing features to my taste.