Running on custom videos

[aakash26]

Hi Authors,

Is it possible to run code on custom videos with extrinsic calibration available?

Regards
Aakash

[fcole]

+Zhengqi Li <zl548@cornell.edu> Hi Aakash, yes, it is possible, though if you would like to run the 2-frame model you will also need flow fields and human masks, and unfortunately we don't have a script to run all those results to generate the input to the network (we used FlowNet2 and Mask RCNN). The TUM evaluation data can serve as a template for what you need, though. cc'ing Zhengqi as well in case he has any additional help. Forrester

…

On Sun, Sep 6, 2020 at 5:10 PM Aakash Rajpal ***@***.***> wrote: Hi Authors, Is it possible to run code on custom videos with extrinsic calibration available? Regards Aakash — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#19>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAD4D3TULPK6POCI3LBOWCLSEP3DXANCNFSM4Q45IVIQ> .

[aakash26]

Hi @fcole ,

Thanks for the reply. I ran the TUM evaulation script test_tum.py for full model-reference. However I am unable to see the dense depth maps for the frames or any results except like the one for test_davis.py which I can see stored under /viz_predictions. How can I see the output for TUM frames?
Also is it possible to clarify what's the exact difference between --input=two_view_k and --input=two_view

Regards
Aakash Rajpal

[Sehaba95]

I have prepared a script to run on a custom video input. I have already created a pull request with a script to run the model on a given video input.