nasib-ullah / video-captioning-models-in-Pytorch

A PyTorch implementation of state of the art video captioning models from 2015-2019 on MSVD and MSRVTT datasets.
70 stars 17 forks source link

About motion feature of dataset #6

Closed Jack2Lu closed 10 months ago

Jack2Lu commented 1 year ago

The framework is very nice and I downloaded the motion feature extracted from resnext-101, but found the shape of feature are different for each video. So when I run the code, it goes wrong , image then I wonder if there's something wrong with motion feature of MSVD and how you extract the motion feature. It's important for me, I hope for your help.

nasib-ullah commented 10 months ago

Hi, for motion features with different shape, change accordingly in config.py file https://github.com/nasib-ullah/video-captioning-models-in-Pytorch/blob/main/config.py#L383