Supervised and Weakly Supervised Deep Learning for Segmentation and Counting of Cotton Bolls Using Proximal Imagery

被引:11
|
作者
Adke, Shrinidhi [1 ,2 ]
Li, Changying [1 ,2 ,3 ]
Rasheed, Khaled M. [1 ,3 ]
Maier, Frederick W. [1 ]
机构
[1] Univ Georgia, Inst Artificial Intelligence, Athens, GA 30602 USA
[2] Univ Georgia, Biosensing & Instrumentat Lab, Coll Engn, Athens, GA 30602 USA
[3] Univ Georgia, Phen & Plant Robot Ctr, Athens, GA 30602 USA
基金
美国国家科学基金会;
关键词
cotton phenotyping; boll counting; supervised learning; mask R-CNN; weakly supervised learning;
D O I
10.3390/s22103688
中图分类号
O65 [分析化学];
学科分类号
070302 ; 081704 ;
摘要
The total boll count from a plant is one of the most important phenotypic traits for cotton breeding and is also an important factor for growers to estimate the final yield. With the recent advances in deep learning, many supervised learning approaches have been implemented to perform phenotypic trait measurement from images for various crops, but few studies have been conducted to count cotton bolls from field images. Supervised learning models require a vast number of annotated images for training, which has become a bottleneck for machine learning model development. The goal of this study is to develop both fully supervised and weakly supervised deep learning models to segment and count cotton bolls from proximal imagery. A total of 290 RGB images of cotton plants from both potted (indoor and outdoor) and in-field settings were taken by consumer-grade cameras and the raw images were divided into 4350 image tiles for further model training and testing. Two supervised models (Mask R-CNN and S-Count) and two weakly supervised approaches (WS-Count and CountSeg) were compared in terms of boll count accuracy and annotation costs. The results revealed that the weakly supervised counting approaches performed well with RMSE values of 1.826 and 1.284 for WS-Count and CountSeg, respectively, whereas the fully supervised models achieve RMSE values of 1.181 and 1.175 for S-Count and Mask R-CNN, respectively, when the number of bolls in an image patch is less than 10. In terms of data annotation costs, the weakly supervised approaches were at least 10 times more cost efficient than the supervised approach for boll counting. In the future, the deep learning models developed in this study can be extended to other plant organs, such as main stalks, nodes, and primary and secondary branches. Both the supervised and weakly supervised deep learning models for boll counting with low-cost RGB images can be used by cotton breeders, physiologists, and growers alike to improve crop breeding and yield estimation.
引用
收藏
页数:17
相关论文
共 50 条
  • [1] Weakly Supervised Deep Learning for Segmentation of Remote Sensing Imagery
    Wang, Sherrie
    Chen, William
    Xie, Sang Michael
    Azzari, George
    Lobell, David B.
    REMOTE SENSING, 2020, 12 (02)
  • [2] A Weakly Supervised Deep Learning Semantic Segmentation Framework
    Zhang, Jizhi
    Zhang, Guoying
    Wang, Qiangyu
    Bai, Shuang
    2017 IEEE INTERNATIONAL CONFERENCE ON SMART CLOUD (SMARTCLOUD), 2017, : 182 - 185
  • [3] Weakly Supervised Semantic Segmentation Based on Deep Learning
    Liang, Binxiu
    Liu, Yan
    He, Linxi
    Li, Jiangyun
    PROCEEDINGS OF THE 11TH INTERNATIONAL CONFERENCE ON MODELLING, IDENTIFICATION AND CONTROL (ICMIC2019), 2020, 582 : 455 - 464
  • [4] Weakly Supervised Instance Segmentation by Deep Community Learning
    Hwang, Jaedong
    Kim, Seohyun
    Son, Jeany
    Han, Bohyung
    2021 IEEE WINTER CONFERENCE ON APPLICATIONS OF COMPUTER VISION (WACV 2021), 2021, : 1019 - 1028
  • [5] A WEAKLY SUPERVISED DEEP LEARNING APPROACH FOR PLANT CENTER DETECTION AND COUNTING
    Karami, Azam
    Crawford, Melba
    Delp, Edward J.
    IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2020, : 1584 - 1587
  • [6] A Weakly Supervised Deep Learning Framework for Sorghum Head Detection and Counting
    Ghosal, Sambuddha
    Zheng, Bangyou
    Chapman, Scott C.
    Potgieter, Andries B.
    Jordan, David R.
    Wang, Xuemin
    Singh, Asheesh K.
    Singh, Arti
    Hirafuji, Masayuki
    Ninomiya, Seishi
    Ganapathysubramanian, Baskar
    Sarkar, Soumik
    Guo, Wei
    PLANT PHENOMICS, 2019, 2019
  • [7] Point in: Counting Trees with Weakly Supervised Segmentation Network
    Tong, Pinmo
    Zhang, Xishan
    Han, Pengcheng
    Bu, Shuhui
    2020 25TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR), 2021, : 9545 - 9552
  • [8] Semantic Segmentation Using a GAN and a Weakly Supervised Method Based on Deep Transfer Learning
    Wen, Shuhuan
    Tian, Wenbo
    Zhang, Hong
    Fan, Shaokang
    Zhou, Nannan
    Li, Xiongfei
    IEEE ACCESS, 2020, 8 : 176480 - 176494
  • [9] Weakly Supervised Segmentation by a Deep Geodesic Prior
    Mortazi, Aliasghar
    Khosravan, Naji
    Torigian, Drew A.
    Kurugol, Sila
    Bagci, Ulas
    MACHINE LEARNING IN MEDICAL IMAGING (MLMI 2019), 2019, 11861 : 238 - 246
  • [10] Weakly Supervised Deep Learning in Radiology
    Misera, Leo
    Mueller-Franzes, Gustav
    Truhn, Daniel
    Kather, Jakob Nikolas
    RADIOLOGY, 2024, 312 (01)