Instance Segmentation by Deep Coloring

TL;DR

This paper introduces a simple method to perform instance segmentation by reducing it to semantic segmentation through dynamic coloring, enabling end-to-end training and effective application across various domains.

Contribution

It presents a novel approach that transforms instance segmentation into a semantic segmentation problem using dynamic coloring, simplifying training and inference.

Findings

Effective on Cityscapes benchmark for autonomous driving

Successful in plant phenotyping and microscopy image analysis

Achieves competitive results with simple architecture

Abstract

We propose a new and, arguably, a very simple reduction of instance segmentation to semantic segmentation. This reduction allows to train feed-forward non-recurrent deep instance segmentation systems in an end-to-end fashion using architectures that have been proposed for semantic segmentation. Our approach proceeds by introducing a fixed number of labels (colors) and then dynamically assigning object instances to those labels during training (coloring). A standard semantic segmentation objective is then used to train a network that can color previously unseen images. At test time, individual object instances can be recovered from the output of the trained convolutional network using simple connected component analysis. In the experimental validation, the coloring approach is shown to be capable of solving diverse instance segmentation tasks arising in autonomous driving (the Cityscapes benchmark), plant phenotyping (the CVPPP leaf segmentation challenge), and high-throughput microscopy image analysis. The source code is publicly available: https://github.com/kulikovv/DeepColoring.