Superpixel Sampling Networks
This addresses the need for task-specific superpixel representations in computer vision, enabling better integration with deep learning models.
The paper tackled the problem of integrating superpixel segmentation into end-to-end trainable deep neural networks by developing a differentiable Superpixel Sampling Network (SSN), which outperforms existing algorithms on segmentation benchmarks and improves downstream network performance.
Superpixels provide an efficient low/mid-level representation of image data, which greatly reduces the number of image primitives for subsequent vision tasks. Existing superpixel algorithms are not differentiable, making them difficult to integrate into otherwise end-to-end trainable deep neural networks. We develop a new differentiable model for superpixel sampling that leverages deep networks for learning superpixel segmentation. The resulting "Superpixel Sampling Network" (SSN) is end-to-end trainable, which allows learning task-specific superpixels with flexible loss functions and has fast runtime. Extensive experimental analysis indicates that SSNs not only outperform existing superpixel algorithms on traditional segmentation benchmarks, but can also learn superpixels for other tasks. In addition, SSNs can be easily integrated into downstream deep networks resulting in performance improvements.