search

ColumbiaDVMM / CDC

CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos
68 stars 18 forks source link

About inference speed #21

Closed Qinying-Liu closed 5 years ago

Qinying-Liu commented 5 years ago

It is mentioned that your speed can reached at around 500 fps. I wonder how you compute the speed. Does it include the time of frames reading and pre-processing before input to the CDC becuse it is usually time-consuming? Hope to receive your reply soon.

bityangke commented 5 years ago

I do not think the inference time here include the pre-reprocessing time. Just feed N samples to the model and calculate the average speed (You can use a batch size).

zhengshou commented 5 years ago

It's running time of caffe part so include pre-processing implemented in caffe but not the part such as generating binary input files beforehand.