search

Atze00 / MoViNet-pytorch

MoViNets PyTorch implementation: Mobile Video Networks for Efficient Video Recognition;
MIT License
256 stars 50 forks source link

Modifying for binary classification #26

Open ekuluke opened 2 years ago

ekuluke commented 2 years ago

Hi, trying to solve a binary classification problem with these movinets and was wondering how I would go about creating a MoVinet with some augmentation with the aim of reducing the TFLOPS required and decreasing the inference time

kaethebu commented 2 years ago

Hey, have you solved your binary classification problem? I'm also working with MoViNets and binary classification atm and have a few questions.

patrick620 commented 2 years ago

Hi, using (2+1)D convolution can reduce computation compared with 3D convolution. But I think the stream model should be well designed to decrease the inference time. Also, the stream buffer is returned by the model and sent back to the model in TF version, but that is stored inside the model in this PyTorch implementation. Maybe TF version is a good choice?!