Adaptive Superpixel for Active Learning in Semantic Segmentation

被引:2
|
作者
Kim, Hoyoung [1 ]
Oh, Minhyeon [2 ]
Hwang, Sehyun [2 ]
Kwak, Suha [1 ,2 ]
Ok, Jungseul [1 ,2 ]
机构
[1] POSTECH, Grad Sch AI, Pohang, South Korea
[2] POSTECH, Dept CSE, Pohang, South Korea
关键词
D O I
10.1109/ICCV51070.2023.00093
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Learning semantic segmentation requires pixel-wise annotations, which can be time-consuming and expensive. To reduce the annotation cost, we propose a superpixel-based active learning (AL) framework, which collects a dominant label per superpixel instead. To be specific, it consists of adaptive superpixel and sieving mechanisms, fully dedicated to AL. At each round of AL, we adaptively merge neighboring pixels of similar learned features into superpixels. We then query a selected subset of these superpixels using an acquisition function assuming no uniform superpixel size. This approach is more efficient than existing methods, which rely only on innate features such as RGB color and assume uniform superpixel sizes. Obtaining a dominant label per superpixel drastically reduces annotators' burden as it requires fewer clicks. However, it inevitably introduces noisy annotations due to mismatches between superpixel and ground truth segmentation. To address this issue, we further devise a sieving mechanism that identifies and excludes potentially noisy annotations from learning. Our experiments on both Cityscapes and PASCAL VOC datasets demonstrate the efficacy of adaptive superpixel and sieving mechanisms.
引用
收藏
页码:943 / 953
页数:11
相关论文
共 50 条
  • [1] Superpixel Transformers for Efficient Semantic Segmentation
    Zhu, Alex Zihao
    Mei, Jieru
    Qiao, Siyuan
    Yan, Hang
    Zhu, Yukun
    Chen, Liang-Chieh
    Kretzschmar, Henrik
    [J]. 2023 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2023, : 7651 - 7658
  • [2] Bayesian Adaptive Superpixel Segmentation
    Uziel, Roy
    Ronen, Meitar
    Freifeld, Oren
    [J]. 2019 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2019), 2019, : 8469 - 8478
  • [3] Adaptive Region Selection for Active Learning in Whole Slide Image Semantic Segmentation
    Qiu, Jingna
    Wilm, Frauke
    Oettl, Mathias
    Schlereth, Maja
    Liu, Chang
    Heimann, Tobias
    Aubreville, Marc
    Breininger, Katharina
    [J]. MEDICAL IMAGE COMPUTING AND COMPUTER ASSISTED INTERVENTION, MICCAI 2023, PT II, 2023, 14221 : 90 - 100
  • [4] Uncertainty-weighted prototype active learning in domain adaptive semantic segmentation
    Dong, Zihao
    Niu, Sijie
    Gao, Xizhan
    Li, Jinping
    Shao, Xiuli
    [J]. EXPERT SYSTEMS WITH APPLICATIONS, 2024, 245
  • [5] Semantic-Aware Superpixel for Weakly Supervised Semantic Segmentation
    Kim, Sangtae
    Park, Daeyoung
    Shim, Byonghyo
    [J]. THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 1, 2023, : 1142 - 1150
  • [6] Superpixel Coherency and Uncertainty Models for Semantic Segmentation
    Baek, SeungRyul
    Lim, Taegyu
    Heo, Yong Seok
    Park, Sungbum
    Kwak, Hantak
    Shim, Woosung
    [J]. 2013 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION WORKSHOPS (ICCVW), 2013, : 275 - 282
  • [7] Hyperspectral Image Classification With Multi-Attention Transformer and Adaptive Superpixel Segmentation-Based Active Learning
    Zhao, Chunhui
    Qin, Boao
    Feng, Shou
    Zhu, Wenxiang
    Sun, Weiwei
    Li, Wei
    Jia, Xiuping
    [J]. IEEE TRANSACTIONS ON IMAGE PROCESSING, 2023, 32 : 3606 - 3621
  • [8] WEAKLY SUPERVISED SEMANTIC SEGMENTATION WITH SUPERPIXEL EMBEDDING
    Xing, Frank Z.
    Cambria, Erik
    Huang, Win-Bin
    Xu, Yang
    [J]. 2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), 2016, : 1269 - 1273
  • [9] Foreground Detection Based on Superpixel and Semantic Segmentation
    Feng, Junying
    Liu, Peng
    Kim, Yong Kwan
    [J]. COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE, 2022, 2022
  • [10] Abstract: Adaptive Region Selection for Active Learning in Whole Slide Image Semantic Segmentation
    Qiu, Jingna
    Wilm, Frauke
    Oettl, Mathias
    Schlereth, Maja
    Liu, Chang
    Heimann, Tobias
    Aubreville, Marc
    Breininger, Katharina
    [J]. BILDVERARBEITUNG FUR DIE MEDIZIN 2024, 2024, : 11 - 11