search

nchucvml / STVT

Video Summarization With Spatiotemporal Vision Transformer
Apache License 2.0
17 stars 7 forks source link

visualization #3

Closed ruanzhijian closed 3 months ago

ruanzhijian commented 3 months ago

Hi author, I would like to know what data your visualization results are depicted by, which variables are available in the code, and also if I want to do a visual analysis of importance scores, how do I do it?

nchucvml commented 3 months ago

The visualization results are based on the important scores are generated by the video summary generation in the code. The scores are directly compared with the ground truth with respect to frame order.