OR-Skip-Net: Outer residual skip network for skin segmentation in non-ideal situations
作者：
Highlights：
• OR-Skip-Net is an end-to-end semantic segmentation network for skin & gland segmentation.
• OR-Skip-Net converges faster using only 9718,786 parameters and keeping edge information.
• OR-Skip-Net uniquely uses the outer residual skip paths from the encoder to directly decoder.
• OR-Skip-Net is tested for skin segmentation of black skin people (BSP).
• Trained OR-Skip-Net, algorithms, and BSP label information are made publicly available.
摘要
•OR-Skip-Net is an end-to-end semantic segmentation network for skin & gland segmentation.•OR-Skip-Net converges faster using only 9718,786 parameters and keeping edge information.•OR-Skip-Net uniquely uses the outer residual skip paths from the encoder to directly decoder.•OR-Skip-Net is tested for skin segmentation of black skin people (BSP).•Trained OR-Skip-Net, algorithms, and BSP label information are made publicly available.
论文关键词：Skin detection,Non-ideal conditions,Hand segmentation,Black skin,OR-Skip-Net
论文评审过程：Received 25 February 2019, Revised 2 September 2019, Accepted 3 September 2019, Available online 4 September 2019, Version of Record 17 September 2019.
论文官网地址：https://doi.org/10.1016/j.eswa.2019.112922
