hou-yz / MVDeTr

[ACM MM 2021] Multiview Detection with Shadow Transformer (and View-Coherent Data Augmentation)
44 stars 8 forks source link

Questions about visualization? #8

Open 131413ljk opened 2 months ago

131413ljk commented 2 months ago

Hi, thanks for the great work on multi-view detection model! I would like to ask how the video visualization video_visualize operation of MV-DETR is like. I found that it is very different from MVDET. How can I implement it? Looking forward to your reply

hou-yz commented 1 month ago

Hi,

Thank you for your interest. As it stands, video_visualize.py will go through all detected pedestrian locations and visualize them on each of the camera views.

Feel free to include more frames to produce a video.

Best, Yunzhong

131413ljk commented 3 weeks ago

Thank you for your reply! This is very good work. If I want to implement it on my own dataset, do you have any good suggestions?