Dmitry Frumkin

Implementation detail of the robust loss

4

Looking at https://gist.github.com/ranftlr/1d6194db2e1dffa0a50c9b0a9549cbd2#file-loss-py-L31, it looks like you are computing the median over all the values, including invalid ones. Shouldn't this be a masked median? (It is not available off the...

Plans to publish the code / dataset?

3

Hello! You have a very interesting paper, and we would be happy to evaluate the results. When do you plan on publishing the evaluation code / training code / dataset?

Validation

1

I see that you have validation turned off in your training script: https://github.com/alibaba/ClickSEG/blob/d61a76b2c88c56106e97d505ce41d16aa2535021/isegm/engine/baseline_trainer.py#L125 Moreover, the code would not work because of https://github.com/alibaba/ClickSEG/blob/d61a76b2c88c56106e97d505ce41d16aa2535021/isegm/data/transforms.py#L8 This should be ``` from albumentations.augmentations.geometric import functional...

Format of crop_size

1

Is crop_size in the [height, width] format or [width, height]? Here it's [height, width]: https://github.com/alibaba/ClickSEG/blob/main/models/strongbaseline/mobilenetv2_x1_comb.py#L41 However, https://github.com/alibaba/ClickSEG/blob/d61a76b2c88c56106e97d505ce41d16aa2535021/isegm/data/aligned_augmentation.py#L50 uses opencv's convention of [width, height] Practically, this does not matter because it's...

Prediction threshold

1

I see you use 0.49 as the prediction threshold. Why this number? https://github.com/alibaba/ClickSEG/blob/d61a76b2c88c56106e97d505ce41d16aa2535021/isegm/engine/baseline_trainer.py#L401

What dataset did you use?

3

Hello Dennis! What dataset did you use for training?