Learning Object-Action Relations from Bimanual Human Demonstration Using Graph Networks

TL;DR

This paper introduces a graph network-based system for frame-wise action classification and segmentation of bimanual human demonstrations using RGB-D data, enabling detailed recognition of simultaneous actions without prior segmentation.

Contribution

It presents a novel approach that extracts symbolic object relations from RGB-D data and trains a graph network classifier for per-frame bimanual action recognition.

Findings

Achieved a macro F1-score of 0.86 in action classification.

Successfully identified true actions within top 3 predictions per frame.

Operates without prior temporal segmentation of actions.

Abstract

Recognizing human actions is a vital task for a humanoid robot, especially in domains like programming by demonstration. Previous approaches on action recognition primarily focused on the overall prevalent action being executed, but we argue that bimanual human motion cannot always be described sufficiently with a single action label. We present a system for frame-wise action classification and segmentation in bimanual human demonstrations. The system extracts symbolic spatial object relations from raw RGB-D video data captured from the robot's point of view in order to build graph-based scene representations. To learn object-action relations, a graph network classifier is trained using these representations together with ground truth action labels to predict the action executed by each hand. We evaluated the proposed classifier on a new RGB-D video dataset showing daily action sequences focusing on bimanual manipulation actions. It consists of 6 subjects performing 9 tasks with 10 repetitions each, which leads to 540 video recordings with 2 hours and 18 minutes total playtime and per-hand ground truth action labels for each frame. We show that the classifier is able to reliably identify (action classification macro F1-score of 0.86) the true executed action of each hand within its top 3 predictions on a frame-by-frame basis without prior temporal action segmentation.