LVIS Challenge Track Technical Report 1st Place Solution: Distribution Balanced and Boundary Refinement for Large Vocabulary Instance Segmentation

TL;DR

This paper presents a novel instance segmentation approach for the LVIS challenge that addresses long-tail distribution and boundary quality, achieving state-of-the-art results through a combination of advanced models, data balancing, and refinement techniques.

Contribution

The paper introduces a distribution balanced method and boundary refinement techniques integrated with a transformer-based backbone to improve large vocabulary instance segmentation.

Findings

Achieved over 45.4% boundary AP on LVIS validation set.

Ranked 1st in LVIS Challenge 2021 with 48.1% AP.

Early stopping and EMA significantly improved segmentation performance.

Abstract

This report introduces the technical details of the team FuXi-Fresher for LVIS Challenge 2021. Our method focuses on the problem in following two aspects: the long-tail distribution and the segmentation quality of mask and boundary. Based on the advanced HTC instance segmentation algorithm, we connect transformer backbone(Swin-L) through composite connections inspired by CBNetv2 to enhance the baseline results. To alleviate the problem of long-tail distribution, we design a Distribution Balanced method which includes dataset balanced and loss function balaced modules. Further, we use a Mask and Boundary Refinement method composed with mask scoring and refine-mask algorithms to improve the segmentation quality. In addition, we are pleasantly surprised to find that early stopping combined with EMA method can achieve a great improvement. Finally, by using multi-scale testing and increasing the upper limit of the number of objects detected per image, we achieved more than 45.4% boundary AP on the val set of LVIS Challenge 2021. On the test data of LVIS Challenge 2021, we rank 1st and achieve 48.1% AP. Notably, our APr 47.5% is very closed to the APf 48.0%.