Not Able To Follow The Codes

[adriannunez]

Not Able To Follow The Codes

[sumeetssingh]

Hello,
Thanks a lot for updating comments. Now its very clear.However, it will be very much helpful if you could also add the script for testing the code using a single video for actual testing. I mean the inference part using any unseen video of fall/no fall. Moreover, I found the URFD dataset contains image sequences not actual videos, but the optical flow scripts which you have mentioned in the paper, accepts videos for generating OF images and the RGB images. Could you please explain a bit how to pass images sequences to the optical flow generator script.

Thanking You

[AdrianNunez]

Hi,

thanks for the interest in the paper and the code. I will be including comments in the code and adding a README file with all the necessary steps to follow in the next days.

Regards.

[AdrianNunez]

Hi,

I will add a function to do inference over just one video (still, it will require optical flow images already processed). And regarding the problem with the URFD dataset, I will upload the modified script I used for processing image sequences instead of videos.

Thanks for the feedback!

[sumeetssingh]

Okay. Thank You Adrian.

[AdrianNunez]

Hi,

Sorry for the delay, I have forked and modified the original repository of the tool to create optical flow images so that it now allows to create them with a sequence of images. Here you have the link to the new repository: https://github.com/AdrianNunez/dense_flow. I have tested it with the URFD dataset without problems.

Regards.