Khawar Islam comments

Results 85 comments of


                                            Khawar Islam

can you provide your pre-trained weight?

@forever208 Is it possible? If we use timm pre-trained weights in this code. The problem is that training takes alot of time.

can you provide your pre-trained weight?

@abhigoku10 Yes, y not https://github.com/rwightman/pytorch-image-models

Train result on my own dataset. A Big gap between Train and Valid dataset

Does anyone train Swin transformers with different image sizes? Like my image size is 112x112, it never works on this size

Model doesn't converge

Hi All, Very good question. I have implemented 8 transformers and out of 7 VITs have a serious convergence problem. The dataset size is 97000 Regards, Khawar

Model doesn't converge

Hi @eslambakr The best model is Praymid ViT Swin, CAIT, CrossViT, CeiT, ViT,NesT, PVT.

Model doesn't converge

@eslambakr I am talking about below paper https://arxiv.org/abs/2102.12122

SwinTransformer and Compact-Transformers Code

Hi @lucidrains you are right about compact transformer but a compact transformer is very efficient in training. They already sent you a Pull request. If possible for you, please add...

SwinTransformer and Compact-Transformers Code

@stevenwalton that's great thank you

How we can set window image size on 112x112 image

@Nial4 I am also facing the same for size 112 but still not working. Any advice?

Bad image quality using 'ffmpeg -c:v libopenjpeg'

I am encoding Kodak images with different sampling factor but my below command gives an error on "sampling-factor" for i in {01..24..1}; do echo JP2 Encoding test/images/kodim$i.png mkdir -p test/JP2/kodim$i...