search

pliang279 / MultiBench

[NeurIPS 2021] Multiscale Benchmarks for Multimodal Representation Learning
MIT License
462 stars 68 forks source link

Code to obtain features from raw data #24

Closed srb-cv closed 2 years ago

srb-cv commented 2 years ago

Could you please tell us if scripts are available to obtain features from the raw video and audio data? Could you point us to the code or provide us with the code?

Vanvan2017 commented 2 years ago

Hey, you can refer to OpenFace and Facet, note that Facet is a commercial tool. So sorry that these features are processed years ago thus it is hard to find the script, if you need all the features you can use the MultimodalSDK, but if you want to deal with some new dataset created by yourself, I think you can try some state-of-the-art method for video and audio especially some fine-tuning ways as you said in the last issue!