BiSeNet mean IoU for R18 about torchseg HOT 20 OPEN

Personally I think you should use OHEM loss for training which his paper didn't mention.

from torchseg.

I get a similar mean IoU 69.688% with R18. I train the model from scratch because the pretrained R18 model is not released. Can you share the pretrained model?

from torchseg.

Personally I think you should use OHEM loss for training which his paper didn't mention.

Thank you but I've already utilized OHEM.

from torchseg.

Maybe you can try large cropsize

from torchseg.

@msc-rajesh Which experiments did you use, cityscapes.bisenet.R18.speed or cityscapes.bisenet.R18? I will re-run it to check the performance.

from torchseg.

@ycszen I run cityscapes.bisenet.R18 with one GPU. I get mean IoU 69.688%.

from torchseg.

@daodaofr I have re-run the cityscapes.bisenet.R18 experiment. The performance is normal. I run this experiment on 4 GPUs. Besides, I think maybe you train from scratch resulting in the performance drop. You can load the official R18 model in Pytorch before I release the pre-trained model.

from torchseg.

@daodaofr : I ran cityscape.bisenet.R18 experiment on 4 GPU's having product name NVidia GeForce GTX 1080 Ti.
I have trained BiSeNet from scratch on CityScape leftImg8bit folder (used "labelTrainIds" instead of "labelIds" from gtFine folder for GT's) with train input image dimensions 1024x1024.
I got mean IoU value 70.446% on validation folder of the CityScape dataset.

from torchseg.

@ycszen @ms-krajesh
I run the model from official R18 model in Pytorch, and I got 72.753% mean IoU.
I think gap is from the pretrained model.

from torchseg.

@daodaofr : Did you used "labelTrainIds" or "labelIds" for GT's?

from torchseg.

@ms-krajesh I used labelIds

from torchseg.

@daodaofr but the class number in labelids is 33, which is not corresponding with the code, do you have any other processes?

from torchseg.

@chenxiaoyu523 Yes, I set the label of invalid classes to ignore_index.

from torchseg.

@ycszen I run your R18.speed and R18 with 4 1080Ti. the accuracies for last epoch models are 74.2 and 75.2. which is a little bit less than your reported accuracy (74.6 and 76.3). Where should be the problem? shall I check the acc for each epoch?

from torchseg.

Can someone upload a pretrained model please ?

from torchseg.

I can't get the right result, and the loss not converge with Bisenet. I don't know why ,can you give me some Suggest

from torchseg.

@jiaxue1993 Maybe you can evaluate the models of the last ten epochs.

from torchseg.

@alexanderfrey The pre-trained models have released except the Xception39.

from torchseg.

@haitaobiyao Could you give more details.

from torchseg.

Hi!, I download the pretrained model(R18)and train model from the link you provided,but I am only able to get mean IoU: 65.2% ,Is there any problem with the parameter setting？

from torchseg.