Panoptic-PartFormer: Learning a Unified Model for Panoptic Part Segmentation

TL;DR

This paper introduces Panoptic-PartFormer, a unified end-to-end transformer-based model that simultaneously performs panoptic and part segmentation, achieving state-of-the-art results with reduced computational cost.

Contribution

We propose the first unified transformer architecture for panoptic part segmentation, integrating all tasks into a single end-to-end model with shared computation.

Findings

Achieves state-of-the-art results on Cityscapes PPS and Pascal Context PPS datasets.

Reduces at least 70% GFlops and 50% parameters compared to previous methods.

Improves accuracy by 3.4% with ResNet50 and 10% with Swin Transformer.

Abstract

Panoptic Part Segmentation (PPS) aims to unify panoptic segmentation and part segmentation into one task. Previous work mainly utilizes separated approaches to handle thing, stuff, and part predictions individually without performing any shared computation and task association. In this work, we aim to unify these tasks at the architectural level, designing the first end-to-end unified method named Panoptic-PartFormer. In particular, motivated by the recent progress in Vision Transformer, we model things, stuff, and part as object queries and directly learn to optimize the all three predictions as unified mask prediction and classification problem. We design a decoupled decoder to generate part feature and thing/stuff feature respectively. Then we propose to utilize all the queries and corresponding features to perform reasoning jointly and iteratively. The final mask can be obtained via inner product between queries and the corresponding features. The extensive ablation studies and analysis prove the effectiveness of our framework. Our Panoptic-PartFormer achieves the new state-of-the-art results on both Cityscapes PPS and Pascal Context PPS datasets with at least 70% GFlops and 50% parameters decrease. In particular, we get 3.4% relative improvements with ResNet50 backbone and 10% improvements after adopting Swin Transformer on Pascal Context PPS dataset. To the best of our knowledge, we are the first to solve the PPS problem via \textit{a unified and end-to-end transformer model. Given its effectiveness and conceptual simplicity, we hope our Panoptic-PartFormer can serve as a good baseline and aid future unified research for PPS. Our code and models are available at https://github.com/lxtGH/Panoptic-PartFormer.