PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model

TL;DR

PP-LiteSeg is a new lightweight real-time semantic segmentation model that balances high accuracy with fast inference speed through innovative modules like FLD, UAFM, and SPPM.

Contribution

The paper introduces PP-LiteSeg, featuring a novel lightweight decoder, a unified attention fusion module, and a simple pyramid pooling module for efficient global context aggregation.

Findings

Achieves 72.0% mIoU at 273.6 FPS on Cityscapes

Outperforms previous real-time segmentation methods in accuracy-speed trade-off

Demonstrates effectiveness of proposed modules in reducing computation

Abstract

Real-world applications have high demands for semantic segmentation methods. Although semantic segmentation has made remarkable leap-forwards with deep learning, the performance of real-time methods is not satisfactory. In this work, we propose PP-LiteSeg, a novel lightweight model for the real-time semantic segmentation task. Specifically, we present a Flexible and Lightweight Decoder (FLD) to reduce computation overhead of previous decoder. To strengthen feature representations, we propose a Unified Attention Fusion Module (UAFM), which takes advantage of spatial and channel attention to produce a weight and then fuses the input features with the weight. Moreover, a Simple Pyramid Pooling Module (SPPM) is proposed to aggregate global context with low computation cost. Extensive evaluations demonstrate that PP-LiteSeg achieves a superior trade-off between accuracy and speed compared to other methods. On the Cityscapes test set, PP-LiteSeg achieves 72.0% mIoU/273.6 FPS and 77.5% mIoU/102.6 FPS on NVIDIA GTX 1080Ti. Source code and models are available at PaddleSeg: https://github.com/PaddlePaddle/PaddleSeg.