search

chaoyuaw / pytorch-coviar

Compressed Video Action Recognition
https://www.cs.utexas.edu/~cywu/projects/coviar/
GNU Lesser General Public License v2.1
500 stars 126 forks source link

Extraction of compressed videos. #62

Open luhc15 opened 5 years ago

luhc15 commented 5 years ago

Thanks for you excellent work, I have a question. In this work , motion vectors are 16x16, but in compressed video, motion vector has other format like 4x4, 8x8, why don't use these mvs of small sizes. It should be more precise.

syed-abdul-baqi commented 5 years ago

Size, smaller than a certain level is not very helpful. 16 works best.