Synthetic Data Supervised Salient Object Detection

TL;DR

This paper introduces SODGAN, a novel method that generates high-quality synthetic image-mask pairs for salient object detection, reducing the need for extensive labeled data and achieving state-of-the-art results.

Contribution

SODGAN is the first approach to generate synthetic data for SOD, utilizing a diffusion embedding network, few-shot mask generator, and quality discriminator to improve training with minimal labeled data.

Findings

Synthetic data achieves 98.4% of the performance of real data-trained models.

Outperforms several fully-supervised state-of-the-art methods.

Excels in semi/weakly-supervised settings.

Abstract

Although deep salient object detection (SOD) has achieved remarkable progress, deep SOD models are extremely data-hungry, requiring large-scale pixel-wise annotations to deliver such promising results. In this paper, we propose a novel yet effective method for SOD, coined SODGAN, which can generate infinite high-quality image-mask pairs requiring only a few labeled data, and these synthesized pairs can replace the human-labeled DUTS-TR to train any off-the-shelf SOD model. Its contribution is three-fold. 1) Our proposed diffusion embedding network can address the manifold mismatch and is tractable for the latent code generation, better matching with the ImageNet latent space. 2) For the first time, our proposed few-shot saliency mask generator can synthesize infinite accurate image synchronized saliency masks with a few labeled data. 3) Our proposed quality-aware discriminator can select highquality synthesized image-mask pairs from noisy synthetic data pool, improving the quality of synthetic data. For the first time, our SODGAN tackles SOD with synthetic data directly generated from the generative model, which opens up a new research paradigm for SOD. Extensive experimental results show that the saliency model trained on synthetic data can achieve $98.4\%$ F-measure of the saliency model trained on the DUTS-TR. Moreover, our approach achieves a new SOTA performance in semi/weakly-supervised methods, and even outperforms several fully-supervised SOTA methods. Code is available at https://github.com/wuzhenyubuaa/SODGAN