Deeply Cascaded U-Net for Multi-Task Image Processing
This addresses the efficiency and performance issues in multi-task image processing for computer vision applications, though it is incremental as it builds on existing U-Net architectures.
The paper tackles the sequential processing of image tasks like denoising and semantic segmentation by proposing a multi-task neural network architecture based on U-Net with additional decoding pathways and deep cascading, achieving better performance and lower trainable parameters compared to individual or jointly-trained networks.
In current practice, many image processing tasks are done sequentially (e.g. denoising, dehazing, followed by semantic segmentation). In this paper, we propose a novel multi-task neural network architecture designed for combining sequential image processing tasks. We extend U-Net by additional decoding pathways for each individual task, and explore deep cascading of outputs and connectivity from one pathway to another. We demonstrate effectiveness of the proposed approach on denoising and semantic segmentation, as well as on progressive coarse-to-fine semantic segmentation, and achieve better performance than multiple individual or jointly-trained networks, with lower number of trainable parameters.