search

fpv-iplab / rulstm

Code for the Paper: Antonino Furnari and Giovanni Maria Farinella. What Would You Expect? Anticipating Egocentric Actions with Rolling-Unrolling LSTMs and Modality Attention. International Conference on Computer Vision, 2019.
http://iplab.dmi.unict.it/rulstm
133 stars 33 forks source link

trying to extract feature using detect_video.py #18

Closed sanketsans closed 2 years ago

sanketsans commented 2 years ago

Hi, I am trying to extract obj features using the detect_video.py script. I installed detectron using this link. But when I run the script, it gives me error of : AssertionError: Models that require precomputed proposals are not supported.

Can you help fix this ?

Thanks

sanketsans commented 2 years ago

I also download pre-extracted detections from the link mentioned earlier, but are they wrt to original video size ie; 1920x1080, or are they already converted to resize video ie; 454x256

antoninofurnari commented 2 years ago

Hello,

Regarding your first question, are you sure you are using the provided config file?. The assertion error you get seems to be related to this line of code, which is executed when cfg.TEST.PRECOMPUTED_PROPOSALS is set to True. Indeed, this should not be set to true as the provided model is Faster RCNN, which does not require pre-computed proposals.

Regarding your second question, we have detected objects at the original resolution.

Please let me know if there are still issues and in that case, please provide more details or the full output of the script, so that we can try to solve it together.

Best, Antonino

sanketsans commented 2 years ago

@antoninofurnari thanks it is working. I am closing this issue.