search

verlab / StraightToThePoint_CVPR_2020

Original PyTorch implementation of the code for the paper "Straight to the Point: Fast-forwarding Videos via Reinforcement Learning Using Textual Data" at the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2020
GNU General Public License v3.0
8 stars 2 forks source link

How to test on different videos? #1

Closed iamgmujtaba closed 3 years ago

iamgmujtaba commented 3 years ago

Hi There,

Thank you for sharing this code. Can you please guide me, how I can test FFMethod on different videos? As per my understanding, I have to prepare a document with sentences. But I could not understand the process relevant to document preparation. What if those sentences/recipe texts are not available in the video?

If possible, please share your thought about document preparation.

Many thanks,

washingtonsk8 commented 3 years ago

Hi,

Unfortunately, at this point of the project, you won't be able to run the fast-forward method without texts associated with it. However, if you know the title of the recipe, you can search for the instructions/directions on websites like All Recipes, WikiHow, or even in a dataset like Recipe1M+. You can also find other videos and their recipes on the WatchNCook website.

Thank you for sharing your doubts,

iamgmujtaba commented 3 years ago

Thank you so much for your explanation.