clovaai/ext_portrait_segmentation

visualizing results

Opened this issue · 4 comments

ucb-pb commented

Setting --visualize to True doesn't appear to do anything.

HYOJINPARK commented

Setting --visualize to True doesn't appear to do anything.

Do you mean that the visdom not working?
when you run : python -m visdom.server ???

ucb-pb commented

Sorry, I mean for running Test_model.py

HYOJINPARK commented

Oh yes I did not add the visualization code in Test_model.py

You can edit by referring the main.py
like this (line 487)

if args.visualize:
            if train_config["loss"] == "Lovasz":
                grid_outputs = torchvision.utils.make_grid(color_transform((save_est[0] > 0).cpu().data), nrow=6)
            else:
                grid_outputs = torchvision.utils.make_grid(color_transform(save_est[0].unsqueeze(0).cpu().max(1)[1].data), nrow=6)

            my_logger.image_summary(torchvision.utils.make_grid(save_input[0], normalize=True),
                             opts = dict(title=f'VAL img (epoch: {epoch})',caption=f'VAL img (epoch: {epoch})'))
            my_logger.image_summary(grid_outputs,
                                      opts=dict(title=f'VAL output (epoch: {epoch}, step: {str(mIOU_val)})',
                                                caption=f'VAL output (epoch: {epoch}, step: {str(mIOU_val)})', ))

            grid_gt = torchvision.utils.make_grid((100 * save_gt[0].cpu()).type('torch.ByteTensor').data,
                                                  nrow=6)
            my_logger.image_summary(grid_gt,
                                      opts=dict(title=f'VAL gt (epoch: {epoch}, step: {str(mIOU_val)})',
                                                caption=f'VAL gt (epoch: {epoch}, step: {str(mIOU_val)})', ))
  • 👍1
HYOJINPARK commented
  • 👍1