1st Place Solution for the 5th LSVOS Challenge: Video Instance Segmentation

TL;DR

This paper presents an improved video instance segmentation method, DVIS, which incorporates denoising training and foundation models, achieving state-of-the-art results and winning the 5th LSVOS Challenge.

Contribution

The paper introduces a denoising training strategy and leverages a frozen visual foundation model to enhance video instance segmentation performance.

Findings

Achieved 57.9 AP on development set

Achieved 56.0 AP on test set

Ranked 1st in the LSVOS Challenge

Abstract

Video instance segmentation is a challenging task that serves as the cornerstone of numerous downstream applications, including video editing and autonomous driving. In this report, we present further improvements to the SOTA VIS method, DVIS. First, we introduce a denoising training strategy for the trainable tracker, allowing it to achieve more stable and accurate object tracking in complex and long videos. Additionally, we explore the role of visual foundation models in video instance segmentation. By utilizing a frozen VIT-L model pre-trained by DINO v2, DVIS demonstrates remarkable performance improvements. With these enhancements, our method achieves 57.9 AP and 56.0 AP in the development and test phases, respectively, and ultimately ranked 1st in the VIS track of the 5th LSVOS Challenge. The code will be available at https://github.com/zhang-tao-whu/DVIS.