search

arthurdouillard / CVPR2021_PLOP

Official code of CVPR 2021's PLOP: Learning without Forgetting for Continual Semantic Segmentation
https://arxiv.org/abs/2011.11390
MIT License
140 stars 23 forks source link

TensorFlow version? #22

Closed andreped closed 2 years ago

andreped commented 2 years ago

Very interesting work! Some of the ideas you propose seem perfect for our applications and we are very interested in testing them :]

However, we are mostly using TF-Keras for ML, so I was wondering if you had an alternative implementation in this framework, or if anyone is trying to reimplement this code to this framework?

arthurdouillard commented 2 years ago

Hey, thank you for your interest in our work 😀

Sorry but I don't have a TF implem, nor know anyone in the Continual Segmentation field coding in TF. A lot of the codebase in that domain are modifications from https://github.com/fcdl94/MiB, which is in PyTorch.

I don't want to start a pytorch vs tensorflow debate, but you may want to consider using pytorch :p

If you find a copy of PLOP in TF, or code one yourself, please share it with me so I can add it in the README.

andreped commented 2 years ago

I don't want to start a pytorch vs tensorflow debate, but you may want to consider using pytorch :p

@arthurdouillard I am using both frameworks myself, but for some applications I just find using Keras so much easier to use (at least for doing simpler stuff - then doing all the low-level stuff necessary with pytorch feels overkill). However with the introduction of pytorch-lightning, I might be on the verge of choosing pytorch as my default framework. Not entirely convinced yet ;)

We might make an attempt to implement parts of PLOP in TF. I will let you know if we do so and we will definitely make it open for others that might be interested.

Thanks for the rapid reply! I am closing this issue for now as my question was answered :]