Dataset Diffusion: Diffusion-based Synthetic Dataset Generation for Pixel-Level Semantic Segmentation

TL;DR

This paper introduces a novel method using Stable Diffusion to generate pixel-level semantic segmentation labels from synthetic images, reducing manual annotation effort and improving segmentation training.

Contribution

The paper presents three innovative techniques leveraging Stable Diffusion's attention mechanisms to produce accurate segmentation maps from text prompts, enabling synthetic dataset creation for semantic segmentation.

Findings

Outperforms existing methods on PASCAL VOC and MSCOCO datasets.

Effectively incorporates uncertainty regions to handle pseudo-label imperfections.

Provides publicly available code and benchmarks for reproducibility.

Abstract

Preparing training data for deep vision models is a labor-intensive task. To address this, generative models have emerged as an effective solution for generating synthetic data. While current generative models produce image-level category labels, we propose a novel method for generating pixel-level semantic segmentation labels using the text-to-image generative model Stable Diffusion (SD). By utilizing the text prompts, cross-attention, and self-attention of SD, we introduce three new techniques: class-prompt appending, class-prompt cross-attention, and self-attention exponentiation. These techniques enable us to generate segmentation maps corresponding to synthetic images. These maps serve as pseudo-labels for training semantic segmenters, eliminating the need for labor-intensive pixel-wise annotation. To account for the imperfections in our pseudo-labels, we incorporate uncertainty regions into the segmentation, allowing us to disregard loss from those regions. We conduct evaluations on two datasets, PASCAL VOC and MSCOCO, and our approach significantly outperforms concurrent work. Our benchmarks and code will be released at https://github.com/VinAIResearch/Dataset-Diffusion