How to display visualization results ？

[iris0329]

I have run the eval_previous_mask.py of one-shot-model-davis model, but how to display visualization results, such as mask or something? I figure out in the /rvos/src/args.py there are some args about visualization and logging, how to use them ? Thanks for your help !

[carlesventura]

I suppose that you have downloaded the one-shot-model-davis and you have extract this zip file into a folder named one-shot-model-davis within the models folder.

If you have run eval_previous_mask.py with the default parameters as given in the scripts, a new folder within the one-shot-model-davis folder should be created containing the binary masks predicted for the different object instances and the different frames.

If you want a more fancy visualization, like the ones included in the paper, then you can use the flag --overlay_masks so another folder will be also created with the RGB images and the masks with some transparency over it.

[iris0329]

Yes, I have downloaded the 'one-shot-model-davis' model and extract it in the right place.

and I use the following parameters:

-model_name=one-shot-model-davis
-dataset=davis2017
-eval_split=val
-batch_size=1
-length_clip=130
-gpu_id=0
--overlay_masks
--visdom
--log_ter

then it surely creates 2 folders as following, but these 2 folders are both empty

Seems wierd. Thanks for your help !

[carlesventura]

Your output in the terminal should be like the one that I attach in this message.

You should see the frame names for each sequence while they are processed. If it is not the case, I would say that you should check the DAVIS dataloader or the LMDB generator (see dataset_lmdb_generator.py).

I hope it helps!

[iris0329]

Thanks for help. I found the problem is I downloaded another version DAVIS2017 after change -eval_split=val to -eval_split=test-dev, it works ! Thank you !