How is the bandwidth of Attentive Fusion (backbone is VoxelNet) 576.71 calculated before compression？

DerrickXuNu / OpenCOOD

[ICRA 2022] An opensource framework for cooperative detection. Official implementation for OPV2V.

https://mobility-lab.seas.ucla.edu/opv2v/

Other

663 stars 99 forks source link

How is the bandwidth of Attentive Fusion (backbone is VoxelNet) 576.71 calculated before compression？ #38

Closed zllxot closed 2 years ago

zllxot commented 2 years ago

Hi！I learned from issue #29 that the bandwidth is calculated by H×W×C×4/100,0000. Then I run voxel_net_intermediate.py and debug to see that the shape of the feature before compression is [5, 128, 200, 704]： but 200 ×704×128×4/100,0000≈72.08 instead of 576.71, what is the problem

DerrickXuNu commented 2 years ago

72.08MB= 576Mb

zllxot commented 2 years ago

72.08MB= 576Mb

ok, thanks for your help.