Improving Real-Time CNN-Based Pupil Detection Through Domain-Specific Data Augmentation

被引:21
|
作者
Eivazi, Shaharam [1 ]
Santini, Thiago [1 ]
Keshavarzi, Alireza [1 ]
Kuebler, Thomas [1 ]
Mazzei, Andrea [2 ]
机构
[1] Univ Tubingen, Percept Engn, Tubingen, Germany
[2] Cort Arts GmbH, Zurich, Switzerland
关键词
Pupil detection; Data augmentation; Deep;
D O I
10.1145/3314111.3319914
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Deep learning is a promising technique for real-world pupil detection. However, the small amount of available accurately-annotated data poses a challenge when training such networks. Here, we utilize non-challenging eye videos where algorithmic approaches perform virtually without errors to automatically generate a foundational data set containing subpixel pupil annotations. Then, we propose multiple domain-specific data augmentation methods to create unique training sets containing controlled distributions of pupil-detection challenges. The feasibility, convenience, and advantage of this approach is demonstrated by training a CNN with these datasets. The resulting network outperformed current methods in multiple publicly-available, realistic, and challenging datasets, despite being trained solely with the augmented eye images. This network also exhibited better generalization w.r.t. the latest state-of-the-art CNN: Whereas on datasets similar to training data, the nets displayed similar performance, on datasets unseen to both networks, ours outperformed the state-of-the-art by approximate to 27% in terms of detection rate.
引用
收藏
页数:6
相关论文
共 50 条
  • [1] Effectiveness of Data Augmentation for CNN-based Pupil Center Point Detection
    Kan, Naoyuki
    Kondo, Nagisa
    Chinsatit, Warapon
    Saitoh, Takeshi
    [J]. 2018 57TH ANNUAL CONFERENCE OF THE SOCIETY OF INSTRUMENT AND CONTROL ENGINEERS OF JAPAN (SICE), 2018, : 41 - 46
  • [2] Improving Domain-Specific NER in the Indonesian Language Through Domain Transfer and Data Augmentation
    Khairunnisa, Siti Oryza
    Chen, Zhousi
    Komachi, Mamoru
    [J]. Journal of Advanced Computational Intelligence and Intelligent Informatics, 2024, 28 (06) : 1299 - 1312
  • [3] Real-Time CNN-Based Driver Distraction & Drowsiness Detection System
    Almazroi, Abdulwahab Ali
    Alqarni, Mohammed A.
    Aslam, Nida
    Shah, Rizwan Ali
    [J]. INTELLIGENT AUTOMATION AND SOFT COMPUTING, 2023, 37 (02): : 2153 - 2174
  • [4] DATA AUGMENTATION FOR CNN-BASED PEOPLE DETECTION IN AERIAL IMAGES
    Chen, Hua-Tsung
    Liu, Che-Han
    Tsai, Wen-Jiin
    [J]. 2018 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA & EXPO WORKSHOPS (ICMEW 2018), 2018,
  • [5] Improving CNN-based activity recognition by data augmentation and transfer learning
    Kalouris, Gerasimos
    Zacharaki, Evangelia I.
    Megalooikonomou, Vasileios
    [J]. 2019 IEEE 17TH INTERNATIONAL CONFERENCE ON INDUSTRIAL INFORMATICS (INDIN), 2019, : 1387 - 1394
  • [6] A CNN-Based Advertisement Recommendation through Real-Time User Face Recognition
    Kim, Gihwi
    Choi, Ilyoung
    Li, Qinglong
    Kim, Jaekyeong
    [J]. APPLIED SCIENCES-BASEL, 2021, 11 (20):
  • [7] Digital Architecture for Real-Time CNN-based Face Detection for Video Processing
    Bhattarai, Smrity
    Madanayake, Arjuna
    Cintra, Renato J.
    Duffner, Stefan
    Garcia, Christophe
    [J]. 2017 COGNITIVE COMMUNICATIONS FOR AEROSPACE APPLICATIONS WORKSHOP (CCAA), 2017,
  • [8] ENHANCING THE QUALITY OF CNN-BASED BURNED AREA DETECTION IN SATELLITE IMAGERY THROUGH DATA AUGMENTATION
    Hnatushenko, Vik.
    Hnatushenko, V.
    Soldatenko, D.
    Heipke, C.
    [J]. GEOSPATIAL WEEK 2023, VOL. 48-1, 2023, : 1749 - 1755
  • [9] CNN-based Note Onset Detection using Synthetic Data Augmentation
    Mounir, Mina
    Karsmakers, Peter
    van Waterschoot, Toon
    [J]. 28TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2020), 2021, : 171 - 175
  • [10] A CNN-Based Approach for Driver Drowsiness Detection by Real-Time Eye State Identification
    Florez, Ruben
    Palomino-Quispe, Facundo
    Coaquira-Castillo, Roger Jesus
    Herrera-Levano, Julio Cesar
    Paixao, Thuanne
    Alvarez, Ana Beatriz
    [J]. APPLIED SCIENCES-BASEL, 2023, 13 (13):