Learning to Count Leaves in Rosette Plants

By Mario Valerio Giuffrida, Massimo Minervini, Sotirios A. Tsaftaris. 2015

110 citations recorded by the source corpus.

Counting the number of leaves in plants is important for plant phenotyping, since it can be used to assess plant growth stages. We propose a learning-based approach for counting leaves in rosette (model) plants. We relate image-based descriptors learned in an unsupervised fashion to leaf counts using a supervised regression model. To take advantage of the circular and coplanar arrangement of leaves and also to introduce scale and rotation invariance, we learn features in a log-polar representation. Image patches extracted in this log-polar domain are provided to K-means, which builds a codebook in a unsupervised manner. Feature codes are obtained by projecting patches on the codebook using the triangle encoding, introducing both sparsity and specifically designed representation. A global, per-plant image descriptor is obtained by pooling local features in specific regions of the image. Finally, we provide the global descriptors to a support vector regression framework to estimate the number of leaves in a plant. We evaluate our method on datasets of the \textit{Leaf Counting Challenge} (LCC), containing images of Arabidopsis and tobacco plants. Experimental results show that on average we reduce absolute counting error by 40% w.r.t. the winner of the 2014 edition of the challenge -a counting via segmentation method. When compared to state-of-the-art density-based approaches to counting, on Arabidopsis image data ~75% less counting errors are observed. Our findings suggest that it is possible to treat leaf counting as a regression problem, requiring as input only the total leaf count per training image.

Subfields: Computer Vision and Pattern Recognition, Pharmacology, Plant Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Pharmacology

In Plant Science

Citations in the corpus

Recurrent Neural Networks for Semantic Instance Segmentation (2017, neutral): “We compare with participants to the CVPPP challenge [39, 42, 30, 15], the two recurrent models in [36, 34] and the metric learning based approach from [12].”
Recurrent Instance Segmentation (2015, background): “– PRIAn [46]: First, a set of features is learned in an unsupervised way on a log-polar representation of the image.”
Finely-grained annotated datasets for image-based plant phenotyping (2016, methodology): “…contests (see Section 4) using a subset of the annotated data presented here showed that the number of available images and criteria used allowed computer vision scientists to work successfully on these problems (see e.g. (Scharr et al., 2015; Pape and Klukas, 2014, 2015; Giu ﬀ rida et al., 2015)).”
Semantic Instance Segmentation with a Discriminative Loss Function (2017, neutral): “6 PRIAn [15] - 1 .”
End-to-End Instance Segmentation with Recurrent Attention (2016, unknown): “8) PRIAn [15] 1.”
segmentation in plant phenotyping: a collation study. Machine Vision and Applications (2017, background): “Interestingly, some of our ﬁndings on learning to count leaves do show that leaf count can be estimated without segmentation [22].”
An interactive tool for semi-automated leaf annotation (2015, methodology): “As an example to demonstrate this fact, we use a learning to count leaves application with regression [13].”
Deep Count: Fruit Counting Based on Deep Simulated Learning (2017, unknown): “.”
Deep Plant Phenomics: A Deep Learning Platform for Complex Plant Phenotyping Tasks (2017, result): “Both results from the literature show signiﬁcantly degraded performance on a selection the three benchmark tasks—multi-accession Arabidopsis for Giuﬀrida et al. (2015), and both Col-0 Arabidopsis and tobacco for Pape and Klukas (2015).”
Leaf segmentation in plant phenotyping: a collation study (2015, background): “Interestingly, some of our findings on learning to count leaves do show that leaf count can be estimated without segmentation [22].”