search

zhihou7 / HOI-CL

Series of work (ECCV2020, CVPR2021, CVPR2021, ECCV2022) about Compositional Learning for Human-Object Interaction Exploration
https://sites.google.com/view/hoi-cl
MIT License
78 stars 11 forks source link

Training ATL on new image set #9

Closed anjugopinath closed 3 years ago

anjugopinath commented 3 years ago

If I want to train the affordance transfer learning code on a new dataset, how many training images should I have for:

i) Human-object interaction ii) Objects

zhihou7 commented 3 years ago

Well, I did not try this and thus I do not know how many training images should it require. We evaluate ATL on V-COCO dataset, which includes 5000 HOI images. Thus, I think 5000 HOI images are enough. But I do not know how many images it at least requires. There are no requirements on the number of Object images since we can utilize the HOI images as object images.

anjugopinath commented 3 years ago

Thank You!