SODAR: Segmenting Objects by DynamicallyAggregating Neighboring Mask Representations

TL;DR

SODAR enhances one-stage instance segmentation by aggregating neighboring mask representations through a learned, efficient method, significantly improving performance with minimal additional computation.

Contribution

The paper introduces SODAR, a novel aggregation approach that leverages neighboring mask information to improve SOLO-based instance segmentation.

Findings

SODAR outperforms SOLO with ResNet-101 by 2.2 AP on COCO test set.

The aggregation method reduces computation by sharing mask representations.

SODAR achieves consistent gains with SOLOv2.

Abstract

Recent state-of-the-art one-stage instance segmentation model SOLO divides the input image into a grid and directly predicts per grid cell object masks with fully-convolutional networks, yielding comparably good performance as traditional two-stage Mask R-CNN yet enjoying much simpler architecture and higher efficiency. We observe SOLO generates similar masks for an object at nearby grid cells, and these neighboring predictions can complement each other as some may better segment certain object part, most of which are however directly discarded by non-maximum-suppression. Motivated by the observed gap, we develop a novel learning-based aggregation method that improves upon SOLO by leveraging the rich neighboring information while maintaining the architectural efficiency. The resulting model is named SODAR. Unlike the original per grid cell object masks, SODAR is implicitly supervised to learn mask representations that encode geometric structure of nearby objects and complement adjacent representations with context. The aggregation method further includes two novel designs: 1) a mask interpolation mechanism that enables the model to generate much fewer mask representations by sharing neighboring representations among nearby grid cells, and thus saves computation and memory; 2) a deformable neighbour sampling mechanism that allows the model to adaptively adjust neighbor sampling locations thus gathering mask representations with more relevant context and achieving higher performance. SODAR significantly improves the instance segmentation performance, e.g., it outperforms a SOLO model with ResNet-101 backbone by 2.2 AP on COCO \texttt{test} set, with only about 3\% additional computation. We further show consistent performance gain with the SOLOv2 model.