# Instance Segmentation by Jointly Optimizing Spatial Embeddings and Clustering Bandwidth

## TL;DR

This paper introduces a novel clustering loss for proposal-free instance segmentation that jointly optimizes spatial embeddings and clustering bandwidth, enabling real-time high-accuracy segmentation suitable for autonomous driving.

## Contribution

It proposes a new loss function and learning approach that improves the speed and accuracy of proposal-free instance segmentation methods.

## Key findings

- Achieves top results on Cityscapes benchmark
- Improves accuracy by 5% over Mask R-CNN
- Operates at over 10 fps on 2MP images

## Abstract

Current state-of-the-art instance segmentation methods are not suited for real-time applications like autonomous driving, which require fast execution times at high accuracy. Although the currently dominant proposal-based methods have high accuracy, they are slow and generate masks at a fixed and low resolution. Proposal-free methods, by contrast, can generate masks at high resolution and are often faster, but fail to reach the same accuracy as the proposal-based methods. In this work we propose a new clustering loss function for proposal-free instance segmentation. The loss function pulls the spatial embeddings of pixels belonging to the same instance together and jointly learns an instance-specific clustering bandwidth, maximizing the intersection-over-union of the resulting instance mask. When combined with a fast architecture, the network can perform instance segmentation in real-time while maintaining a high accuracy. We evaluate our method on the challenging Cityscapes benchmark and achieve top results (5\% improvement over Mask R-CNN) at more than 10 fps on 2MP images. Code will be available at https://github.com/davyneven/SpatialEmbeddings .