DeepPod: a convolutional neural network based quantification of fruit number in Arabidopsis.

Azam Hamidinekoo,Chuan Lu,Fiona M K Corke,Gina A Garzón-Martínez,John H Doonan,Morteza Ghahremani,Reyer Zwiggelaar

doi:10.1093/gigascience/giaa012

Abstract

BackgroundHigh-throughput phenotyping based on non-destructive imaging has great potential in plant biology and breeding programs. However, efficient feature extraction and quantification from image data remains a bottleneck that needs to be addressed. Advances in sensor technology have led to the increasing use of imaging to monitor and measure a range of plants including the model Arabidopsis thaliana. These extensive datasets contain diverse trait information, but feature extraction is often still implemented using approaches requiring substantial manual input.ResultsThe computational detection and segmentation of individual fruits from images is a challenging task, for which we have developed DeepPod, a patch-based 2-phase deep learning framework. The associated manual annotation task is simple and cost-effective without the need for detailed segmentation or bounding boxes. Convolutional neural networks (CNNs) are used for classifying different parts of the plant inflorescence, including the tip, base, and body of the siliques and the stem inflorescence. In a post-processing step, different parts of the same silique are joined together for silique detection and localization, whilst taking into account possible overlapping among the siliques. The proposed framework is further validated on a separate test dataset of 2,408 images. Comparisons of the CNN-based prediction with manual counting (R2 = 0.90) showed the desired capability of methods for estimating silique number.ConclusionsThe DeepPod framework provides a rapid and accurate estimate of fruit number in a model system widely used by biologists to investigate many fundemental processes underlying growth and reproduction

Highlights

Photometrics promises to revolutionize biological research and agricultural production systems [1,2,3,4,5]
A convolutional neural network (CNN) contains convolutional layers, where inputs are passed through various filters for extracting features that are arranged as feature maps, prior to using the fully connected layers for classification or regression
A CNN can be trained using a back-propagation algorithm to update the weights in an iterative process, in order to minimize the loss function that measures the discrepancy between the predicted output and actual output for the training examples

Summary

Introduction

Photometrics (imaging followed by computationally assisted feature extraction and measurement) promises to revolutionize biological research and agricultural production systems [1,2,3,4,5]. High-throughput phenotyping technologies aim to address this problem by the use of non-destructive approaches either. Advances in sensor technology have led to the increasing use of imaging to monitor and measure a range of plants including the model Arabidopsis thaliana. These extensive datasets contain diverse trait information, but feature extraction is often still implemented using approaches requiring substantial manual input. Results: The computational detection and segmentation of individual fruits from images is a challenging task, for which we have developed DeepPod, a patch-based 2-phase deep learning framework. Convolutional neural networks (CNNs) are used for classifying different parts of the plant inflorescence, including the tip, base, and body of the siliques and the stem inflorescence. What makes CNNs attractive in computer vision is that they can directly extract features from images without the need for time-consuming, hand-crafted pre-processing or feature extraction steps, unlike classical machine learning approaches [34]

Results

Discussion

Conclusion