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.
|Title of host publication||Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings|
|Editors||Vittorio Ferrari, Cristian Sminchisescu, Martial Hebert, Yair Weiss|
|Number of pages||18|
|Publication status||Published - 2018|
|Event||15th European Conference on Computer Vision, ECCV 2018 - Munich, Germany|
Duration: 2018 Sep 8 → 2018 Sep 14
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Other||15th European Conference on Computer Vision, ECCV 2018|
|Period||18/9/8 → 18/9/14|
Bibliographical notePublisher Copyright:
© Springer Nature Switzerland AG 2018.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Computer Science(all)