open-solution-mapping-challenge

open-solution-mapping-challenge copied to clipboard

hickle = h5py + pickle

1

Official implementation of the [Fully Convolutional Instance-aware Semantic Segmentation](https://arxiv.org/abs/1611.07709) (FCIS) based on MXNet -> they use hickle to dump masks fast. It is wotrh checking if it is useful to...

kamil-kaczmarek

kamil-kaczmarek

* use datasets from the same organization (their open data-hackathons).

jakubczakon

* Currently normalization constants `mean=0` `std=1` are hard-coded, they should be calculated on train and passed to loaders * Substitute `mean` and `std` with values from pretrained PyTorch models (i.e....

jakubczakon

Use Pillow-SIMD for better performance

1

Pillow can be replaced by a faster Pillow-SIMD with little to no effort. I think it is worth trying https://github.com/uploadcare/pillow-simd

jakubczakon