search

Wangt-CN / VC-R-CNN

[CVPR 2020] The official pytorch implementation of ``Visual Commonsense R-CNN''
MIT License
352 stars 61 forks source link

Up_Down_VC downstream task #10

Closed windspirit95 closed 4 years ago

windspirit95 commented 4 years ago

Hi, Thank you for such a great open-source here. I have succeeded in making an inference to the raw image base on ruotianluo 's codebase. I am also interest on using your VC Feature there, so I guess it will work with VC concatenated with UpDown Feature on MSCOCO dataset. About raw data inference, it is mentioned by ruotianluo that it will not work with UpDown Feature. Therefore, is it possible to concatenate your VC Feature with the pre-trained ResNet features, or we should use another mechanism? Thank you very much.

Wangt-CN commented 4 years ago

Actually the raw image can not work with updown feature is because the image should be firstly extracted the Updown feature. You can just get the updown feature firstly. About concatenating to the pretrained resnet feature, the answer is absolutely yes. But I haven't done the experiment.

Wangt-CN commented 4 years ago

For the updown feature, you can refer to here

Wangt-CN commented 4 years ago

Moreover, the feature concatenation actually is very free. You can just treat VC feature as an additional information source and design models to use it.

windspirit95 commented 4 years ago

Thank you, I will try it and let you know soon ^^