Solving Current Limitations of Deep Learning Based Approaches for Plant Disease Detection

Arsenović, Marko; Karanovic, Mirjana; Sladojević, Srdjan; Anderla, Andraš; Stefanović, Darko

doi:10.3390/sym11070939

Cited by 309 publications

(129 citation statements)

References 44 publications

Supporting

Mentioning

127

Contrasting

Unclassified

Order By: Relevance

“…This way, images of leaves with different shapes, angles, levels of infection, main leaf color, brightness, ambient lighting, etc., are all included in all categories in order to achieve the highest possible variability. Variability in the datasets ensures that the model will be trained in the most generalized fashion possible, and will be evaluated and tested under all conditions [25].…”

Section: Data Splitmentioning

confidence: 99%

A Convolutional Neural Networks Based Method for Anthracnose Infected Walnut Tree Leaves Identification

et al. 2020

View full text Add to dashboard Cite

Anthracnose is a fungal disease that infects a large number of trees worldwide, damages intensively the canopy, and spreads with ease to neighboring trees, resulting in the potential destruction of whole crops. Even though it can be treated relatively easily with good sanitation, proper pruning and copper spraying, the main issue is the early detection for the prevention of spreading. Machine learning algorithms can offer the tools for the on-site classification of healthy and affected leaves, as an initial step towards managing such diseases. The purpose of this study was to build a robust convolutional neural network (CNN) model that is able to classify images of leaves, depending on whether or not these are infected by anthracnose, and therefore determine whether a tree is infected. A set of images were used both in grayscale and RGB mode, a fast Fourier transform was implemented for feature extraction, and a CNN architecture was selected based on its performance. Finally, the best performing method was compared with state-of-the-art convolutional neural network architectures.

show abstract

Section: Data Splitmentioning

confidence: 99%

A Convolutional Neural Networks Based Method for Anthracnose Infected Walnut Tree Leaves Identification

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Authors in [ 7 ] proposed a two-stage algorithm tackling the problem of plant disease detection in complex background environments and severe environmental conditions. The authors claimed to have collected the largest image dataset of around 80 thousand images of 12 different species with 42 distinct classes.…”

Section: Literature Reviewmentioning

confidence: 99%

“…In order to assist farmers in disease detection, different techniques based on artificial intelligence and computer vision have been proposed to identify the disease along with its severity level accurately. However, most of the experiments are performed in a controlled environment [ 7 ]. Essential factors such as image angle, scale, direction, and brightness are often ignored under such conditions.…”

Section: Introductionmentioning

confidence: 99%

Disease Detection in Plum Using Convolutional Neural Network under True Field Conditions

Ahmad

Jan

Farman

et al. 2020

Sensors

View full text Add to dashboard Cite

The agriculture sector faces crop losses every year due to diseases around the globe, which adversely affect food productivity and quality. Detecting and identifying plant diseases at an early stage is still a challenge for farmers, particularly in developing countries. Widespread use of mobile computing devices and the advancements in artificial intelligence have created opportunities for developing technologies to assist farmers in plant disease detection and treatment. To this end, deep learning has been widely used for disease detection in plants with highly favorable outcomes. In this paper, we propose an efficient convolutional neural network-based disease detection framework in plum under true field conditions for resource-constrained devices. As opposed to the publicly available datasets, images used in this study were collected in the field by considering important parameters of image-capturing devices such as angle, scale, orientation, and environmental conditions. Furthermore, extensive data augmentation was used to expand the dataset and make it more challenging to enable robust training. Investigations of recent architectures revealed that transfer learning of scale-sensitive models like Inception yield results much better with such challenging datasets with extensive data augmentation. Through parameter quantization, we optimized the Inception-v3 model for deployment on resource-constrained devices. The optimized model successfully classified healthy and diseased fruits and leaves with more than 92% accuracy on mobile devices.

show abstract

“…While Isokane et al 28 used the synthetic plant models for the estimation of branching pattern, Giuffrida et al used GAN-generated images to train a neural network for Arabidopsis leaf counting 29 . Similarly, Arsenovic et al used StyleGAN 30 to create training images for the plant disease image classification 31 . Meanwhile, Ward et al generated artificial images of Arabidopsis rendered from 3D models and utilized them for neural network training in leaf segmentation 32 .…”

Section: Introductionmentioning

confidence: 99%

Learning from Synthetic Dataset for Crop Seed Instance Segmentation

Toda¹,

Okura

Ito

et al. 2019

Preprint

View full text Add to dashboard Cite

12 13 *Correspondence should be addressed to Yosuke Toda; tyosuke@aquaseerser.com 14 15Incorporating deep learning in the image analysis pipeline has opened the possibility of introducing precision 16 phenotyping in the field of agriculture. However, to train the neural network, a sufficient amount of training 17 data must be prepared, which requires a time-consuming manual data annotation process that often becomes 18 the limiting step. Here, we show that an instance segmentation neural network (Mask R-CNN) aimed to 19 phenotype the barley seed morphology of various cultivars, can be sufficiently trained purely by a 20 synthetically generated dataset. Our attempt is based on the concept of domain randomization, where a large 21 amount of image is generated by randomly orienting the seed object to a virtual canvas. After training with 22 such a dataset, performance based on recall and the average Precision of the real-world test dataset achieved 23 96% and 95%, respectively. Applying our pipeline enables extraction of morphological parameters at a large 24 scale, enabling precise characterization of the natural variation of barley from a multivariate perspective. 25Importantly, we show that our approach is effective not only for barley seeds but also for various crops 26 including rice, lettuce, oat, and wheat, and thus supporting the fact that the performance benefits of this 27 technique is generic. We propose that constructing and utilizing such synthetic data can be a powerful method 28 to alleviate human labor costs needed to prepare the training dataset for deep learning in the agricultural 29 domain. 30 31 32

show abstract

Solving Current Limitations of Deep Learning Based Approaches for Plant Disease Detection

Cited by 309 publications

References 44 publications

A Convolutional Neural Networks Based Method for Anthracnose Infected Walnut Tree Leaves Identification

A Convolutional Neural Networks Based Method for Anthracnose Infected Walnut Tree Leaves Identification

Disease Detection in Plum Using Convolutional Neural Network under True Field Conditions

Learning from Synthetic Dataset for Crop Seed Instance Segmentation

Contact Info

Product

Resources

About