Skip to content

Please add Bongard-HOI and RelViT #4

@jeasinema

Description

@jeasinema

Hi,

Thanks for making this learning list and indeed I learned a lot. Just want to share some of our recent work on visual reasoning and I hope they could help this community through your platform:

Bongard-HOI: Benchmarking Few-Shot Visual Reasoning for Human-Object Interactions (CVPR 2022, oral)
arxiv | code
In this work, we create a new dataset for few-shot visual reasoning with HOI concepts. It introduces great challenges to the state-of-the-art few-shot learners but seems straightforward for humans.

RelViT: Concept-guided Vision Transformer for Visual Relational Reasoning (ICLR 2022)
arxiv | code
In this work, we explore vision transformers for many visual relational reasoning tasks, including HICO and GQA. We further introduce concept-guided contrastive learning that helps these models master visual reasoning without massive pertaining or extra training data.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions