Class-Incremental Learning of Plant and Disease Detection: Growing Branches with Knowledge Distillation

被引:0
|
作者
Page-Fortin, Mathieu [1 ]
机构
[1] Laval Univ, Quebec City, PQ, Canada
关键词
D O I
10.1109/ICCVW60793.2023.00066
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper investigates the problem of class-incremental object detection for agricultural applications where a model needs to learn new plant species and diseases incrementally without forgetting the previously learned ones. We adapt two public datasets to include new categories over time, simulating a more realistic and dynamic scenario. We then compare three class-incremental learning methods that leverage different forms of knowledge distillation to mitigate catastrophic forgetting. Our experiments show that all three methods suffer from catastrophic forgetting, but the Dynamic Y-KD approach, which additionally uses a dynamic architecture that grows new branches to learn new tasks, outperforms ILOD and Faster-ILOD in most settings both on new and old classes. These results highlight the challenges and opportunities of continual object detection for agricultural applications. In particular, we hypothesize that the large intra-class and small inter-class variability that is typical of plant images exacerbate the difficulty of learning new categories without interfering with previous knowledge. We publicly release our code to encourage future work.
引用
收藏
页码:593 / 603
页数:11
相关论文
共 50 条
  • [1] Class-Incremental Learning by Knowledge Distillation with Adaptive Feature Consolidation
    Kang, Minsoo
    Park, Jaeyoo
    Han, Bohyung
    [J]. 2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022), 2022, : 16050 - 16059
  • [2] CLASS-INCREMENTAL LEARNING FOR REMOTE SENSING IMAGES BASED ON KNOWLEDGE DISTILLATION
    Song, Jingduo
    Jia, Hecheng
    Xu, Feng
    [J]. IGARSS 2023 - 2023 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2023, : 5026 - 5028
  • [3] Hyperspectral Image Classification Based on Class-Incremental Learning with Knowledge Distillation
    Xu, Meng
    Zhao, Yuanyuan
    Liang, Yajun
    Ma, Xiaorui
    [J]. REMOTE SENSING, 2022, 14 (11)
  • [4] Few-Shot Class-Incremental Learning via Relation Knowledge Distillation
    Dong, Songlin
    Hong, Xiaopeng
    Tao, Xiaoyu
    Chang, Xinyuan
    Wei, Xing
    Gong, Yihong
    [J]. THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2021, 35 : 1255 - 1263
  • [5] AN END-TO-END ARCHITECTURE FOR CLASS-INCREMENTAL OBJECT DETECTION WITH KNOWLEDGE DISTILLATION
    Hao, Yu
    Fu, Yanwei
    Jiang, Yu-Gang
    Tian, Qi
    [J]. 2019 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO (ICME), 2019, : 1 - 6
  • [6] Class-Incremental Exemplar Compression for Class-Incremental Learning
    Luo, Zilin
    Liu, Yaoyao
    Schiele, Bernt
    Sun, Qianru
    [J]. 2023 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2023, : 11371 - 11380
  • [7] Multi-granularity knowledge distillation and prototype consistency regularization for class-incremental learning
    Shi, Yanyan
    Shi, Dianxi
    Qiao, Ziteng
    Wang, Zhen
    Zhang, Yi
    Yang, Shaowu
    Qiu, Chunping
    [J]. NEURAL NETWORKS, 2023, 164 : 617 - 630
  • [8] Semantic Knowledge Guided Class-Incremental Learning
    Wang, Shaokun
    Shi, Weiwei
    Dong, Songlin
    Gao, Xinyuan
    Song, Xiang
    Gong, Yihong
    [J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2023, 33 (10) : 5921 - 5931
  • [9] Class-Incremental Learning on Video-Based Action Recognition by Distillation of Various Knowledge
    Maraghi, Vali Ollah
    Faez, Karim
    [J]. COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE, 2022, 2022
  • [10] Semantic-aware Knowledge Distillation for Few-Shot Class-Incremental Learning
    Cheraghian, Ali
    Rahman, Shafin
    Fang, Pengfei
    Roy, Soumava Kumar
    Petersson, Lars
    Harandi, Mehrtash
    [J]. 2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, : 2534 - 2543