Zhao (Dylan) Wang comments

Results 37 comments of


                                            Zhao (Dylan) Wang

Downstream KUMC test compile issue

@10086ddd KUMC dataset is more difficult:)

Question about metric

Hi @Lycus99 , thanks for your interest! For this metric, we directly adopt it from https://github.com/Beckschen/TransUNet, you can refer to this discussion: https://github.com/Beckschen/TransUNet/issues/39

Question for input

Hi @kimsekeun thanks for your interest! 1. We use 16 frames from each video to pre-train FM. 2. We feed an entire video at once for segmentation training. Our model...

AttributeError: 'NoneType' object has no attribute 'glGetError'

Thanks for your prompt reply! Unfortunately, I can not install with under root permission. I have tried to reinstall pyrender==0.1.33 and pyopengl==3.1.4, but meeting the same issue. Is there any...

Use Endo-FM for laparoscopic video feature extraction

Hi, @carolinafidalgo Thanks for your interest! Can I have a look of your own data?

Downstream Fine-tuning for Segmentation Task

Hi, @KOO-DE thanks for your interest in our work! The pre-trained weight is loaded here: https://github.com/med-air/Endo-FM/blob/3374b57cd4a5f5c3e999270ab916463ca6f49807/TransUNet/networks/vit_seg_modeling.py#L260

Downstream Fine-tuning for Segmentation Task

Hi, @KOO-DE you may load our pre-trained weights `endo_fm.pth` for segmentation fine-tuning. The original TransUNet used the pre-trained weights on ImageNet, which may be not fit in endoscopy scenario due...