1st Place Solution for the UVO Challenge on Image-based Open-World Segmentation 2021

TL;DR

This paper presents a two-stage, class-agnostic instance segmentation framework that combines object detection and segmentation, achieving first place in the UVO 2021 challenge.

Contribution

The novel approach integrates object detection with segmentation in a class-agnostic manner, leading to top performance in open-world segmentation tasks.

Findings

Achieved first place in the UVO 2021 challenge

Effective class-agnostic training of segmentation networks

High-quality object proposals for open-world segmentation

Abstract

We describe our two-stage instance segmentation framework we use to compete in the challenge. The first stage of our framework consists of an object detector, which generates object proposals in the format of bounding boxes. Then, the images and the detected bounding boxes are fed to the second stage, where a segmentation network is applied to segment the objects in the bounding boxes. We train all our networks in a class-agnostic way. Our approach achieves the first place in the UVO 2021 Image-based Open-World Segmentation Challenge.