NIR Iris Challenge Evaluation in Non-cooperative Environments: Segmentation and Localization Caiyong Wang 1,* , Yunlong Wang 2 , Kunbo Zhang 2 , Jawad Muhammad 2 , Tianhao Lu 2 , Qi Zhang 3 , Qichuan Tian 1 , Zhaofeng He 4 , Zhenan Sun 2,† , Yiwen Zhang 5 , Tianbao Liu 5 , Wei Yang 5 , Dongliang Wu 6 , Yingfeng Liu 6 , Ruiye Zhou 6 , Huihai Wu 6 , Hao Zhang 7 , Junbao Wang 7 , Jiayi Wang 7 , Wantong Xiong 7 , Xueyu Shi 8 , Shao Zeng 8 , Peihua Li 8 , Haodong Sun 9 , Jing Wang 9 , Jiale Zhang 9 , Qi Wang 9 , Huijie Wu 10 , Xinhui Zhang 10 , Haiqing Li 10 , Yu Chen 11 , Liang Chen 11 , Menghan Zhang 11 , Ye Sun 12 , Zhiyong Zhou 12 , Fadi Boutros 13 , Naser Damer 13 , Arjan Kuijper 13 , Juan Tapia 14 , Andres Valenzuela 14 , Christoph Busch 14 , Gourav Gupta 15 , Kiran Raja 15 , Xi Wu 16 , Xiaojie Li 16 , Jingfu Yang 16 , Hongyan Jing 16 , Xin Wang 17 , Bin Kong 17 , Youbing Yin 17 , Qi Song 17 , Siwei Lyu 18 , Shu Hu 18 , Leon Premk 19 , Matej Vitek 19 , Vitomir ˇ struc 19 , Peter Peer 19 , Jalil Nourmohammadi Khiarak 20 , Farhang Jaryani 21 , Samaneh Salehi Nasab 22 , Seyed Naeim Moafinejad 23 , Yasin Amini 24 , Morteza Noshad 25 1 School of Electrical and Information Engineering, Beijing University of Civil Engineering and Architecture, China, 2 CRIPAC, NLPR, CASIA, China, 3 People’s Public Security University of China, China, 4 Beijing University of Posts and Telecommunications, China, 5 School of Biomedical Engineering, Southern Medical University, China, 6 Shanghai University of Electric Power, China, 7 College of Science, Northeastern University, China, 8 Dalian University of Technology, China, 9 College of Sciences, Northeastern University, China, 10 IriStar Technology Co., Ltd, China, 11 Xi’an Quanxiu Technology Co., Ltd, China, 12 JiLin University, China, 13 Fraunhofer Institute for Computer Graphics Research IGD, Germany; Mathematical and Applied Visual Computing, TU Darmstadt, Germany, 14 Darmstadt University of Applied Sciences (Hochschule Darmstadt) and TOC Biometrics, 15 Norwegian University of Science and Technology, Norway, 16 Chengdu University of Information Technology, China, 17 Keya Medical, Seattle, WA, USA, 18 University at Buffalo, USA, 19 University of Ljubljana, Slovenia, 20 Warsaw University of Technology, Poland, 21 Arak University, Iran, 22 Lorestan University, Iran, 23 Shahid Beheshti University, Iran, 24 University of Kharazmi, Iran, 25 Stanford University, USA * wangcaiyong@bucea.edu.cn, † znsun@nlpr.ia.ac.cn(corresponding author) Abstract For iris recognition in non-cooperative environments, iris segmentation has been regarded as the first most im- portant challenge still open to the biometric community, af- fecting all downstream tasks from normalization to recog- nition. In recent years, deep learning technologies have gained significant popularity among various computer vi- sion tasks and also been introduced in iris biometrics, espe- cially iris segmentation. To investigate recent developments and attract more interest of researchers in the iris segmen- tation method, we organized the 2021 NIR Iris Challenge E- valuation in Non-cooperative Environments: Segmentation and Localization (NIR-ISL 2021) at the 2021 International Joint Conference on Biometrics (IJCB 2021). The challenge was used as a public platform to assess the performance of iris segmentation and localization methods on Asian and African NIR iris images captured in non-cooperative envi- ronments. The three best-performing entries achieved solid and satisfactory iris segmentation and localization results in most cases, and their code and models have been made publicly available for reproducibility research. 1. Introduction Traditional iris recognition usually imposes many con- straints to the user cooperation and imaging conditions, which seriously limits the application range of iris recog- nition. To solve this dilemma, there has been much re- 978-1-6654-3780-6/21/$31.00 c 2021 IEEE cent work [7, 3, 15, 9] on the non-cooperative or less- constrained iris recognition (either at-a-distance, on-the- move, with minor user cooperation, within dynamic imag- ing environments and using mobile devices). Under these circumstances, captured iris images inevitably suffer from all kinds of noise, such as occlusions due to eyelids or eye- lashes, specular reflections, off-angle, or blur. To make full use of these noisy iris images, efficient and robust iris segmentation has been regarded as the first most important challenge still open to the biometric community, affecting all downstream tasks from normalization to recognition. In 2007, Noisy Iris Challenge Evaluation - Part I (NICE.I) was held to benchmark the iris segmentation methods on the Noisy Visible Wavelength Iris Image Database (UBIRIS.v2) [7]. In 2013, Mobile Iris CHallenge Evaluation part I (MICHE I) was also held to evaluate the iris segmentation methods developed for visible iris images captured with mobile devices [3]. The two iris segmentation benchmarking competitions mainly focused on the VIS iris images from Caucasian people. In addition, most submitted methods in NICE.I and MICHE I were developed based on traditional image processing and pattern recognition meth- ods, rather than deep learning technologies emerged in re- cent years [8, 2]. To reflect latest developments of iris segmentation and offer new insights, the 2021 NIR Iris Challenge Evalua- tion in Non-cooperative Environments: Segmentation and Localization (NIR-ISL 2021) was organised as part of the 2021 International Joint Conference on Biometrics (IJCB 2021). The challenge aimed at benchmarking the method- s of iris segmentation and localization for NIR iris images 2021 IEEE International Joint Conference on Biometrics (IJCB) | 978-1-6654-3780-6/21/$31.00 ©2021 IEEE | DOI: 10.1109/IJCB52358.2021.9484336 Authorized licensed use limited to: FhI fur Graphische Datenverarbeitung. Downloaded on August 03,2021 at 09:50:05 UTC from IEEE Xplore. Restrictions apply.