Problem
Research questions and friction points this paper is trying to address.
Image Denoising
Complexity Reduction
Self-supervised Learning
SwinIA: Self-Supervised Blind-Spot Image Denoising without Convolutions
🤖 AI Summary
Self-supervised image denoising without paired ground-truth data remains challenging, especially when avoiding handcrafted priors or auxiliary supervision.
Method: We propose the first purely Transformer-based blind-spot denoising method—eliminating convolutions, pixel masking, explicit noise priors, and complex regularizers. Built upon the Swin Transformer, our end-to-end autoencoder leverages windowed self-attention to inherently satisfy the blind-spot constraint, trained solely with MSE loss.
Contribution/Results: This work establishes the first fully Transformer-driven self-supervised denoiser, achieving state-of-the-art performance across multiple standard benchmarks. The approach is computationally efficient, highly generalizable, and requires neither clean images nor assumptions about noise distribution—significantly enhancing practicality and scalability of blind denoising.
📝 Abstract
Self-supervised image denoising implies restoring the signal from a noisy image without access to the ground truth. State-of-the-art solutions for this task rely on predicting masked pixels with a fully-convolutional neural network. This most often requires multiple forward passes, information about the noise model, or intricate regularization functions. In this paper, we propose a Swin Transformer-based Image Autoencoder (SwinIA), the first fully-transformer architecture for self-supervised denoising. The flexibility of the attention mechanism helps to fulfill the blind-spot property that convolutional counterparts normally approximate. SwinIA can be trained end-to-end with a simple mean squared error loss without masking and does not require any prior knowledge about clean data or noise distribution. Simple to use, SwinIA establishes the state of the art on several common benchmarks.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Swin Transformer
Self-supervised Denoising
Attention Mechanism
🔎 Similar Papers
No similar papers found.
