search

JJBOY / BMN-Boundary-Matching-Network

A pytorch-version implementation codes of paper: "BMN: Boundary-Matching Network for Temporal Action Proposal Generation", which is accepted in ICCV 2019.
MIT License
291 stars 64 forks source link

temporal dimension ？？？ #40

Open Aries5522 opened 3 years ago

Aries5522 commented 3 years ago

this repo run on ActivityNet with feature shape 400100,RGB200100 concat flow 200100 ,do you know how to make feature temporal dimension to 100?when i extract feature on Thumos 14 ,I get feature 4096L,L is length of video frames/16. should i do a meanpooling on temporal dimension to fix value???

JJBOY commented 3 years ago

On THUMOS14, you can use a sliding window to truncate the L to many small clips with length 100. On ActivityNet, we use bilinear interpolation to scale the L to 100. For more details, you can refer to the paper BSN and BMN.

Aries5522 commented 3 years ago

Thanks for reply!!!!!!!and have you train on Thumos14,i use code GTAD data process like sliding window ,it is much difference from ActivityNet,and i get MAP=0.34 when tIOU=0.5, but for long action instance i think it doesnt work , can i know your result on Thumos14?

Aries5522 commented 3 years ago

my wechat is Demo20190325 ,can you add my wechat