FRED-Net: Fully residual encoder–decoder network for accurate iris segmentation

作者：

Highlights：

• Proposed FRED-Net is an end-to-end semantic segmentation network for iris and road scene.

• FRED-Net is a standalone network where eyelid, eyelash, and glint detections are not required.

• FRED-Net is the result of step-by-step development, whose step is a complete variant network.

• FRED-Net uses the residual connectivity for both encoder and decoder.

• Our trained FRED-Net models, along with the algorithms, are made publicly available.

摘要

•Proposed FRED-Net is an end-to-end semantic segmentation network for iris and road scene.•FRED-Net is a standalone network where eyelid, eyelash, and glint detections are not required.•FRED-Net is the result of step-by-step development, whose step is a complete variant network.•FRED-Net uses the residual connectivity for both encoder and decoder.•Our trained FRED-Net models, along with the algorithms, are made publicly available.

论文关键词：Iris recognition,Iris segmentation,Full residual encoder–decoder network,Semantic segmentation

论文评审过程：Received 14 August 2018, Revised 27 December 2018, Accepted 4 January 2019, Available online 7 January 2019, Version of Record 30 January 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.01.010