DragAPart: Learning a Part-Level Motion Prior for Articulated Objects

TL;DR

DragAPart is a novel method that learns part-level motion priors for articulated objects, enabling realistic and category-generalized motion generation from images and user interactions.

Contribution

It introduces DragAPart, a framework that predicts part-level interactions using a pre-trained image generator fine-tuned on a synthetic dataset, enhancing motion understanding across categories.

Findings

Outperforms prior motion-controlled generators in part-level understanding

Generalizes well to real images and multiple object categories

Uses a new encoding for drags and dataset randomization

Abstract

We introduce DragAPart, a method that, given an image and a set of drags as input, generates a new image of the same object that responds to the action of the drags. Differently from prior works that focused on repositioning objects, DragAPart predicts part-level interactions, such as opening and closing a drawer. We study this problem as a proxy for learning a generalist motion model, not restricted to a specific kinematic structure or object category. We start from a pre-trained image generator and fine-tune it on a new synthetic dataset, Drag-a-Move, which we introduce. Combined with a new encoding for the drags and dataset randomization, the model generalizes well to real images and different categories. Compared to prior motion-controlled generators, we demonstrate much better part-level motion understanding.