Lecture 1: Swin Transformer from Scratch in PyTorch - Hierarchic Structure and Shifted Windows Ideas

[@viallykazadimutombo225]

This is the best of bests tutorial I've seen so far about Swin Transformer. I'm very much impressed!

[@ai_hmp]

Thank you so much for your kind comment. :)

[@iumemon]

Hey Bro, I didn't know you were working on Vision Transformer.

[@harusakichiwa]

Very well explained. Thank you for making such a good tutorial 👍

[@ai_hmp]

Thank you.

[@BradNeuberg]

Very good, clear explanation, thanks for making this!

[@ai_hmp]

I am glad you found it helpful :)

[@waqarmughal4755]

Thanks for the excellent videos. I have a question how can I train my custom data using that swin transformers

[@akramsalim9706]

Thank you very much. You are the best

[@ai_hmp]

I am glad you found it helpful :)

[@syedadzha362]

Amazing explaination. Thank you!

[@ai_hmp]

Thank you!

[@SS-zq5sc]

This is perfect, thank you very much! Really well explained.

[@ai_hmp]

Glad it was helpful!

[@ChizkiyahuOhayon]

How can a Swin Transformer backbone output multi-level feature maps like ResNet?

[@malathip4043]

Can this be used on character images of 32*32*1 dataset

[@icanyagmur]

Very intuitive!

[@ai_hmp]

Thank you!

[@purnamakkena9553]

Is there any tutorial like this for understanding self attention? You are really explaining nicely. Thank You

[@ai_hmp]

Thank you so much! Regarding self-attention, I found a course on Coursera that is very easy to understand and well-explained: Introduction to Machine Learning, Duke University, Week 5.

[@Krishnavandejagadguru-bo1ng]

How to trained the swin transformer for the rectangle image i.e 64x512x3

[@ai_hmp]

Thank you for your question. The window is still square but you will end up having less number of windows on height than width. The rest should be similar.

[@smarttaurian30]

what about the segmenatation?

[@ai_hmp]

We only focused on the backbone in this series.

[@RomeoKienzler]

Very nicely explained. Thanks so much!

[@ai_hmp]

Thank you so much :)