Leveraging multiple datasets for deep leaf counting

Preprint English OPEN
Dobrescu, Andrei ; Giuffrida, Mario Valerio ; Tsaftaris, Sotirios A (2017)
  • Subject: Computer Science - Computer Vision and Pattern Recognition

The number of leaves a plant has is one of the key traits (phenotypes) describing its development and growth. Here, we propose an automated, deep learning based approach for counting leaves in model rosette plants. While state-of-the-art results on leaf counting with deep learning methods have recently been reported, they obtain the count as a result of leaf segmentation and thus require per-leaf (instance) segmentation to train the models (a rather strong annotation). Instead, our method treats leaf counting as a direct regression problem and thus only requires as annotation the total leaf count per plant. We argue that combining different datasets when training a deep neural network is beneficial and improves the results of the proposed approach. We evaluate our method on the CVPPP 2017 Leaf Counting Challenge dataset, which contains images of Arabidopsis and tobacco plants. Experimental results show that the proposed method significantly outperforms the winner of the previous CVPPP challenge, improving the results by a minimum of ~50% on each of the test datasets, and can achieve this performance without knowing the experimental origin of the data (i.e. in the wild setting of the challenge). We also compare the counting accuracy of our model with that of per leaf segmentation algorithms, achieving a 20% decrease in mean absolute difference in count (|DiC|).
  • References (32)
    32 references, page 1 of 4

    [1] C. Arteta, V. Lempitsky, J. A. Noble, and A. Zisserman. Interactive Object Counting. pages 504-518, 2014.

    [2] C. Arteta, V. Lempitsky, and A. Zisserman. Counting in the Wild. 1:483-498, 2016.

    [3] J. Bell and H. Dee. Aberystwyth Leaf Evaluation Dataset, 2016.

    [4] A. Chayeb, N. Ouadah, Z. Tobal, M. Lakrouf, and O. Azouaoui. Hog based multi-object detection for urban navigation. In 17th International IEEE Conference on Intelligent Transportation Systems (ITSC), pages 2962-2967, 2014.

    [5] J. A. Cruz, X. Yin, X. Liu, S. M. Imran, D. D. Morris, D. M. Kramer, and J. Chen. Multi-modality imagery database for plant phenotyping. Machine Vision and Applications, 27(5):735-749, 2016.

    [6] N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05), volume 1, pages 886-893 vol. 1, June 2005.

    [7] J. Donahue, L. Anne Hendricks, S. Guadarrama, M. Rohrbach, S. Venugopalan, K. Saenko, and T. Darrell. Long-term recurrent convolutional networks for visual recognition and description. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2015.

    [8] L. Fiaschi, R. Nair, U. Koethe, and F. A. Hamprecht. Learning to Count with Regression Forest and Structured Labels. In 21st International Conference on Pattern Recognition (ICPR 2012), pages 2685-2688, 2012.

    [9] R. Girshick. Fast r-cnn. In The IEEE International Conference on Computer Vision (ICCV), December 2015.

    [10] R. Girshick, J. Donahue, T. Darrell, and J. Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014.

  • Metrics
    No metrics available
Share - Bookmark