All About Knowledge Graphs for Actions

TL;DR

This paper investigates how different types of knowledge graphs affect zero-shot and few-shot action recognition, providing insights and an improved evaluation framework for knowledge transfer in video datasets.

Contribution

It systematically analyzes three knowledge graph construction methods for action recognition and proposes a new evaluation paradigm for zero-shot and few-shot learning.

Findings

Action-object embeddings improve recognition accuracy.

Visual embeddings provide complementary benefits.

The proposed evaluation framework standardizes comparisons.

Abstract

Current action recognition systems require large amounts of training data for recognizing an action. Recent works have explored the paradigm of zero-shot and few-shot learning to learn classifiers for unseen categories or categories with few labels. Following similar paradigms in object recognition, these approaches utilize external sources of knowledge (eg. knowledge graphs from language domains). However, unlike objects, it is unclear what is the best knowledge representation for actions. In this paper, we intend to gain a better understanding of knowledge graphs (KGs) that can be utilized for zero-shot and few-shot action recognition. In particular, we study three different construction mechanisms for KGs: action embeddings, action-object embeddings, visual embeddings. We present extensive analysis of the impact of different KGs in different experimental setups. Finally, to enable a systematic study of zero-shot and few-shot approaches, we propose an improved evaluation paradigm based on UCF101, HMDB51, and Charades datasets for knowledge transfer from models trained on Kinetics.