GHOST: Grounded Human Motion Generation with Open Vocabulary Scene-and-Text Contexts

TL;DR

This paper introduces GHOST, a novel open vocabulary scene encoder for grounded human motion generation that leverages knowledge distillation and regularization to improve accuracy and flexibility in multi-modal context understanding.

Contribution

It proposes a two-step approach with pretraining via knowledge distillation and fine-tuning with novel regularization, enabling better text-scene connection and motion grounding.

Findings

Achieves up to 30% reduction in goal object distance metric.

Demonstrates improved performance over state-of-the-art on HUMANISE dataset.

Framework is adaptable to future segmentation methods.

Abstract

The connection between our 3D surroundings and the descriptive language that characterizes them would be well-suited for localizing and generating human motion in context but for one problem. The complexity introduced by multiple modalities makes capturing this connection challenging with a fixed set of descriptors. Specifically, closed vocabulary scene encoders, which require learning text-scene associations from scratch, have been favored in the literature, often resulting in inaccurate motion grounding. In this paper, we propose a method that integrates an open vocabulary scene encoder into the architecture, establishing a robust connection between text and scene. Our two-step approach starts with pretraining the scene encoder through knowledge distillation from an existing open vocabulary semantic image segmentation model, ensuring a shared text-scene feature space. Subsequently, the scene encoder is fine-tuned for conditional motion generation, incorporating two novel regularization losses that regress the category and size of the goal object. Our methodology achieves up to a 30% reduction in the goal object distance metric compared to the prior state-of-the-art baseline model on the HUMANISE dataset. This improvement is demonstrated through evaluations conducted using three implementations of our framework and a perceptual study. Additionally, our method is designed to seamlessly accommodate future 2D segmentation methods that provide per-pixel text-aligned features for distillation.