AllSpark: Reborn Labeled Features from Unlabeled in Transformer for Semi-Supervised Semantic Segmentation

TL;DR

AllSpark introduces a novel transformer-based module that reuses labeled features from unlabeled data to improve semi-supervised semantic segmentation, achieving superior results without complex training pipelines.

Contribution

The paper proposes AllSpark, a new architecture component that enhances semi-supervised segmentation by rebirthing labeled features from unlabeled data using cross-attention, avoiding complex training schemes.

Findings

Outperforms existing methods on Pascal, Cityscapes, and COCO benchmarks.

Effectively integrates into transformer-based segmentation models.

Achieves state-of-the-art results without complicated training procedures.

Abstract

Semi-supervised semantic segmentation (SSSS) has been proposed to alleviate the burden of time-consuming pixel-level manual labeling, which leverages limited labeled data along with larger amounts of unlabeled data. Current state-of-the-art methods train the labeled data with ground truths and unlabeled data with pseudo labels. However, the two training flows are separate, which allows labeled data to dominate the training process, resulting in low-quality pseudo labels and, consequently, sub-optimal results. To alleviate this issue, we present AllSpark, which reborns the labeled features from unlabeled ones with the channel-wise cross-attention mechanism. We further introduce a Semantic Memory along with a Channel Semantic Grouping strategy to ensure that unlabeled features adequately represent labeled features. The AllSpark shed new light on the architecture level designs of SSSS rather than framework level, which avoids increasingly complicated training pipeline designs. It can also be regarded as a flexible bottleneck module that can be seamlessly integrated into a general transformer-based segmentation model. The proposed AllSpark outperforms existing methods across all evaluation protocols on Pascal, Cityscapes and COCO benchmarks without bells-and-whistles. Code and model weights are available at: https://github.com/xmed-lab/AllSpark.