Compositional Scene Understanding through Inverse Generative Modeling
This work addresses the problem of compositional scene understanding for computer vision, offering incremental improvements by applying inverse modeling to existing generative frameworks.
The paper tackles scene understanding by framing it as an inverse generative modeling problem, where conditional parameters of a compositional generative model are inferred to fit natural images, enabling robust generalization to new scenes with increased objects and shapes, and achieving zero-shot multi-object perception with pretrained models.
Generative models have demonstrated remarkable abilities in generating high-fidelity visual content. In this work, we explore how generative models can further be used not only to synthesize visual content but also to understand the properties of a scene given a natural image. We formulate scene understanding as an inverse generative modeling problem, where we seek to find conditional parameters of a visual generative model to best fit a given natural image. To enable this procedure to infer scene structure from images substantially different than those seen during training, we further propose to build this visual generative model compositionally from smaller models over pieces of a scene. We illustrate how this procedure enables us to infer the set of objects in a scene, enabling robust generalization to new test scenes with an increased number of objects of new shapes. We further illustrate how this enables us to infer global scene factors, likewise enabling robust generalization to new scenes. Finally, we illustrate how this approach can be directly applied to existing pretrained text-to-image generative models for zero-shot multi-object perception. Code and visualizations are at https://energy-based-model.github.io/compositional-inference.