Training Matting Models without Alpha Labels

TL;DR

This paper introduces a novel training method for deep image matting models that uses coarse annotations like trimaps instead of precise alpha labels, leveraging semantic learning and a directional distance consistency loss to propagate alpha values effectively.

Contribution

It proposes a new training paradigm that eliminates the need for fine alpha labels by using rough trimaps and a directional distance consistency loss, enabling effective alpha matte inference.

Findings

Achieves comparable performance to fine-label supervised models.

Sometimes outperforms human-labeled ground truth.

Validates on AM-2K and P3M-10K datasets.

Abstract

The labelling difficulty has been a longstanding problem in deep image matting. To escape from fine labels, this work explores using rough annotations such as trimaps coarsely indicating the foreground/background as supervision. We present that the cooperation between learned semantics from indicated known regions and proper assumed matting rules can help infer alpha values at transition areas. Inspired by the nonlocal principle in traditional image matting, we build a directional distance consistency loss (DDC loss) at each pixel neighborhood to constrain the alpha values conditioned on the input image. DDC loss forces the distance of similar pairs on the alpha matte and on its corresponding image to be consistent. In this way, the alpha values can be propagated from learned known regions to unknown transition areas. With only images and trimaps, a matting model can be trained under the supervision of a known loss and the proposed DDC loss. Experiments on AM-2K and P3M-10K dataset show that our paradigm achieves comparable performance with the fine-label-supervised baseline, while sometimes offers even more satisfying results than human-labelled ground truth. Code is available at \url{https://github.com/poppuppy/alpha-free-matting}.