Recurrent U-Net for Resource-Constrained Segmentation
This addresses the need for efficient and high-performance segmentation models for applications such as medical imaging and autonomous driving, though it appears incremental as it builds upon the U-Net framework.
The paper tackles the problem of resource-constrained segmentation by introducing a recurrent U-Net architecture that outperforms state-of-the-art methods on benchmarks like hand, retina vessel, and road segmentation, while maintaining compactness.
State-of-the-art segmentation methods rely on very deep networks that are not always easy to train without very large training datasets and tend to be relatively slow to run on standard GPUs. In this paper, we introduce a novel recurrent U-Net architecture that preserves the compactness of the original U-Net, while substantially increasing its performance to the point where it outperforms the state of the art on several benchmarks. We will demonstrate its effectiveness for several tasks, including hand segmentation, retina vessel segmentation, and road segmentation. We also introduce a large-scale dataset for hand segmentation.