Deep Plant Phenomics: A Deep Learning Platform for Complex Plant Phenotyping Tasks

Ubbens, Jordan; Stavness, Ian

doi:10.3389/fpls.2017.01190

Cited by 280 publications

(237 citation statements)

References 30 publications

Supporting

Mentioning

231

Contrasting

Unclassified

Order By: Relevance

“…The second is to experiment with a wider range of potential models. For simplicity of comparison, here we adopted a previously proposed CNN model for plant phenomics including a total of four convolutional layers [14]. However, models with more layers and/or more complex model structures could also be evaluated, as these have been shown to provided higher prediction accuracy in some use cases [7,30].…”

Section: Discussionmentioning

confidence: 99%

“…Training of leaf counting models was conducted using the Deep Plant Phenomics (DPP) package, which provides a plant phenotyping specific interface and wrapper on top of TensorFlow (v1.10) [14,39]. Prior to both model training and testing, the input size of each image was set to 256 × 256 pixels.…”

Section: Training and Evaluation Of Cnn Modelsmentioning

confidence: 99%

“…The first set of approaches may require greater investments of skilled labor in designing and troubleshooting algorithms, but smaller sets of ground truth measurements to obtain suitable accuracy of measurement [5,[8][9][10][11]. The second set of approaches, based on deep machine learning, often require substantially larger ground truth datasets to act as training data, but smaller investments of programmer time per additional trait scored [12][13][14][15][16]. The use of training datasets which were either insufficient in size or those with low quality annotations can lead to low prediction accuracy and "over fitting", where training results in excellent prediction accuracy on the training data itself but low prediction when applied to newly collected data.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Simulated Plant Images Improve Maize Leaf Counting Accuracy

Miao

Hoban

Pages

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

Automatically scoring plant traits using a combination of imaging and deep learning holds promise to accelerate data collection, scientific inquiry, and breeding progress. However, applications of this approach are currently held back by the availability of large and suitably annotated training datasets. Early training datasets targeted arabidopsis or tobacco. The morphology of these plants quite different from that of grass species like maize. Two sets of maize training data, one real-world and one synthetic were generated and annotated for late vegetative stage maize plants using leaf count as a model trait. Convolutional neural networks (CNNs) trained on entirely synthetic data provided predictive power for scoring leaf number in real-world images. This power was less than CNNs trained with equal numbers of real-world images, however, in some cases CNNs trained with larger numbers of synthetic images outperformed CNNs trained with smaller numbers of real-world images. When real-world training images were scarce, augmenting real-world training data with synthetic data provided improved prediction accuracy. Quantifying leaf number over time can provide insight into plant growth rates and stress responses, and can help to parameterize crop growth models. The approaches and annotated training data described here may help future efforts to develop accurate leaf counting algorithms for maize.4. Partially or completely overlapping leaves from the perspective of the observer ( Figure 2D).

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Training and Evaluation Of Cnn Modelsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Simulated Plant Images Improve Maize Leaf Counting Accuracy

Miao

Hoban

Pages

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…In recent years machine learning has driven advances throughout many computer vision domains [3]. Indeed, much of the recent progress in plant phenotyping has also been driven by new and so-called deep learning techniques, a branch of AI, often centring around Convolutional Neural Networks (CNNs) [4,5,6]. The sharp increase in the availability of performant techniques in image analysis has coincided with an increase in the availability of genomic information in plant biol-Compiled on: July 19, 2019.…”

Section: Introductionmentioning

confidence: 99%

RootNav 2.0: Deep Learning for Automatic Navigation of Complex Plant Root Architectures

Yasrab

Atkinson

Wells

et al. 2019

Preprint

View full text Add to dashboard Cite

We present a new image analysis approach that provides fully-automatic extraction of complex root system architectures from a range of plant species in varied imaging setups. Driven by modern deep-learning approaches, RootNav 2.0 replaces previously manual and semi-automatic feature extraction with an extremely deep multi-task Convolutional Neural Network architecture. The network has been designed to explicitly combine local pixel information with global scene information in order to accurately segment small root features across high-resolution images. In addition, the network simultaneously locates seeds, and rst and second order root tips to drive a search algorithm seeking optimal paths throughout the image, extracting accurate architectures without user interaction. The proposed method is evaluated on images of wheat (Triticum aestivum L.) from a seedling assay. The results are compared with semi-automatic analysis via the original RootNav tool, demonstrating comparable accuracy, with a 10-fold increase in speed. We then demonstrate the ability of the network to adapt to di erent plant species via transfer learning, o ering similar accuracy when transferred to an Arabidopsis thaliana plate assay. We transfer for a nal time to images of Brassica napus from a hydroponic assay, and still demonstrate good accuracy despite many fewer training images. The tool outputs root architectures in the widely accepted RSML standard, for which numerous analysis packages exist (http://rootsystemml.github.io/), as well as segmentation masks compatible with other automated measurement tools.

show abstract

“…Recently, deep neural networks have also been employed to address the leaf counting problem. These approaches essentially combine the task of finding suitable image features with the task of learning a good regression model relating the features to leaf count (Aich and Stavness, 2017;Dobrescu et al, 2017a;Ubbens and Stavness, 2017). These approaches show significant promise, but each of these is specialized: a new model and network for each plant species or cultivar, imaging condition, etc is required.…”

Section: Introductionmentioning

confidence: 99%

Pheno‐Deep Counter: a unified and versatile deep learning architecture for leaf counting

2018

View full text Add to dashboard Cite

SummaryDirect observation of morphological plant traits is tedious and a bottleneck for high‐throughput phenotyping. Hence, interest in image‐based analysis is increasing, with the requirement for software that can reliably extract plant traits, such as leaf count, preferably across a variety of species and growth conditions. However, current leaf counting methods do not work across species or conditions and therefore may lack broad utility. In this paper, we present Pheno‐Deep Counter, a single deep network that can predict leaf count in two‐dimensional (2D) plant images of different species with a rosette‐shaped appearance. We demonstrate that our architecture can count leaves from multi‐modal 2D images, such as visible light, fluorescence and near‐infrared. Our network design is flexible, allowing for inputs to be added or removed to accommodate new modalities. Furthermore, our architecture can be used as is without requiring dataset‐specific customization of the internal structure of the network, opening its use to new scenarios. Pheno‐Deep Counter is able to produce accurate predictions in many plant species and, once trained, can count leaves in a few seconds. Through our universal and open source approach to deep counting we aim to broaden utilization of machine learning‐based approaches to leaf counting. Our implementation can be downloaded at https://bitbucket.org/tuttoweb/pheno-deep-counter.

show abstract

Deep Plant Phenomics: A Deep Learning Platform for Complex Plant Phenotyping Tasks

Cited by 280 publications

References 30 publications

Simulated Plant Images Improve Maize Leaf Counting Accuracy

Simulated Plant Images Improve Maize Leaf Counting Accuracy

RootNav 2.0: Deep Learning for Automatic Navigation of Complex Plant Root Architectures

Pheno‐Deep Counter: a unified and versatile deep learning architecture for leaf counting

Contact Info

Product

Resources

About