ConsNet: Learning Consistency Graph for Zero-Shot Human-Object Interaction Detection

TL;DR

ConsNet introduces a graph-based approach that leverages multi-level consistencies among objects, actions, and interactions to improve zero-shot and supervised human-object interaction detection.

Contribution

The paper proposes ConsNet, a novel framework that encodes relations among HOI components into a graph and uses GATs to enhance detection, especially for unseen categories.

Findings

Outperforms state-of-the-art on V-COCO and HICO-DET datasets.

Effective in zero-shot HOI detection scenarios.

Utilizes visual and semantic features for improved recognition.

Abstract

We consider the problem of Human-Object Interaction (HOI) Detection, which aims to locate and recognize HOI instances in the form of <human, action, object> in images. Most existing works treat HOIs as individual interaction categories, thus can not handle the problem of long-tail distribution and polysemy of action labels. We argue that multi-level consistencies among objects, actions and interactions are strong cues for generating semantic representations of rare or previously unseen HOIs. Leveraging the compositional and relational peculiarities of HOI labels, we propose ConsNet, a knowledge-aware framework that explicitly encodes the relations among objects, actions and interactions into an undirected graph called consistency graph, and exploits Graph Attention Networks (GATs) to propagate knowledge among HOI categories as well as their constituents. Our model takes visual features of candidate human-object pairs and word embeddings of HOI labels as inputs, maps them into visual-semantic joint embedding space and obtains detection results by measuring their similarities. We extensively evaluate our model on the challenging V-COCO and HICO-DET datasets, and results validate that our approach outperforms state-of-the-arts under both fully-supervised and zero-shot settings. Code is available at https://github.com/yeliudev/ConsNet.