Materialistic: Selecting Similar Materials in Images

TL;DR

Materialistic introduces a robust, similarity-based method for selecting regions of images with the same material, regardless of shading or lighting, using unsupervised features and a new synthetic dataset.

Contribution

The paper proposes a novel approach leveraging unsupervised DINO features and a Cross-Similarity module for material-based image region selection, trained on a synthetic dataset and applicable to real images.

Findings

Method generalizes well to real-world images

Effective in material editing and retrieval applications

Robust to shading, highlights, and shadows

Abstract

Separating an image into meaningful underlying components is a crucial first step for both editing and understanding images. We present a method capable of selecting the regions of a photograph exhibiting the same material as an artist-chosen area. Our proposed approach is robust to shading, specular highlights, and cast shadows, enabling selection in real images. As we do not rely on semantic segmentation (different woods or metal should not be selected together), we formulate the problem as a similarity-based grouping problem based on a user-provided image location. In particular, we propose to leverage the unsupervised DINO features coupled with a proposed Cross-Similarity module and an MLP head to extract material similarities in an image. We train our model on a new synthetic image dataset, that we release. We show that our method generalizes well to real-world images. We carefully analyze our model's behavior on varying material properties and lighting. Additionally, we evaluate it against a hand-annotated benchmark of 50 real photographs. We further demonstrate our model on a set of applications, including material editing, in-video selection, and retrieval of object photographs with similar materials.