mzolfaghari / ECO-efficient-video-understanding

Code and models of paper " ECO: Efficient Convolutional Network for Online Video Understanding", ECCV 2018
MIT License
437 stars 96 forks source link

in the file models_ECO_Lite/kinetics/deploy.prototxt why the input_dim is 80? #27

Closed l2009312042 closed 5 years ago

l2009312042 commented 5 years ago

name: "ECOLite" ####################################### data ####################################### input: "data" input_dim: 80

why the input_dim is 80 ?
is batchsize 5 ?16 * 5 =80 ,is that right ?

mzolfaghari commented 5 years ago

Yes, it's right. You can have any (number*16) depending on your GPU memory.