search

fredzzhang / spatially-conditioned-graphs

[ICCV'21] Official PyTorch implementation for paper "Spatially Conditioned Graphs for Detecting Human–Object Interactions"
https://arxiv.org/abs/2012.06060
BSD 3-Clause "New" or "Revised" License
63 stars 17 forks source link

How can i use the pretrained HICO model for OKVQA action detections #64

Closed ifmaq1 closed 3 years ago

ifmaq1 commented 3 years ago

Hi, I want to retrieve the action detections on OKVQA dataset. I only want to detect actions on OKVQA dataset using pretrained HICO model. Can you please guide me how do I do that?

Also, do I need object detections on OKVQA beforehand? in order to use HICO pretrained model on it?

I read https://github.com/fredzzhang/spatially-conditioned-graphs/issues/63 too. but I didn't understand how to implement it to my problem

fredzzhang commented 3 years ago

Hi @ifmaq1 ,

You do need object detections to run the model. The simplest thing to do is to run Faster R-CNN pre-trained on MS COCO on your dataset. The model is available in torchvision model zoo.

To run our model on custom dataset, you just need to get the inputs correct. The model takes three inputs as shown here, a list of images, a list of detections and a list of targets. If you are running inference only, leave the target as None. The detections should be a list of dictionaries as shown here.

Hope that answers you questions.

Cheers, Fred.

ifmaq1 commented 3 years ago

Thanks alot. That was really helpful