A new benchmark for group distribution shifts in hand grasp regression for object manipulation. Can meta-learning raise the bar?

TL;DR

This paper introduces a new benchmark for evaluating hand and object pose regression under distribution shifts in real-world scenarios, and investigates whether meta-learning can improve adaptation to unknown objects.

Contribution

It proposes a novel benchmark for group distribution shifts in hand-object pose regression and evaluates meta-learning as a method to enhance generalization to unseen objects.

Findings

Meta-learning shows measurable improvements over baseline models.

Optimization interference occurs in joint hand-object pose regression with meta-learning.

Analysis provides insights for future research on this benchmark.

Abstract

Understanding hand-object pose with computer vision opens the door to new applications in mixed reality, assisted living or human-robot interaction. Most methods are trained and evaluated on balanced datasets. This is of limited use in real-world applications; how do these methods perform in the wild on unknown objects? We propose a novel benchmark for object group distribution shifts in hand and object pose regression. We then test the hypothesis that meta-learning a baseline pose regression neural network can adapt to these shifts and generalize better to unknown objects. Our results show measurable improvements over the baseline, depending on the amount of prior knowledge. For the task of joint hand-object pose regression, we observe optimization interference for the meta-learner. To address this issue and improve the method further, we provide a comprehensive analysis which should serve as a basis for future work on this benchmark.