A mobile-based deep learning model for cassava disease diagnosis

Convolutional neural network (CNN) models have the potential to improve plant disease phenotyping where the standard approach is visual diagnostics requiring specialized training. In scenarios where a CNN is deployed on mobile devices, models are presented with new challenges due to lighting and ori...

Descripción completa

Detalles Bibliográficos
Autores principales: Ramcharan, A., McCloskey, P., Baranowski, K., Mbilinyi, N., Mrisho, L., Ndalahwa, M., Legg, James P., Hughes, D.P.
Formato: Journal Article
Lenguaje:Inglés
Publicado: Frontiers Media 2019
Materias:
Acceso en línea:https://hdl.handle.net/10568/105535
_version_ 1855515665943756800
author Ramcharan, A.
McCloskey, P.
Baranowski, K.
Mbilinyi, N.
Mrisho, L.
Ndalahwa, M.
Legg, James P.
Hughes, D.P.
author_browse Baranowski, K.
Hughes, D.P.
Legg, James P.
Mbilinyi, N.
McCloskey, P.
Mrisho, L.
Ndalahwa, M.
Ramcharan, A.
author_facet Ramcharan, A.
McCloskey, P.
Baranowski, K.
Mbilinyi, N.
Mrisho, L.
Ndalahwa, M.
Legg, James P.
Hughes, D.P.
author_sort Ramcharan, A.
collection Repository of Agricultural Research Outputs (CGSpace)
description Convolutional neural network (CNN) models have the potential to improve plant disease phenotyping where the standard approach is visual diagnostics requiring specialized training. In scenarios where a CNN is deployed on mobile devices, models are presented with new challenges due to lighting and orientation. It is essential for model assessment to be conducted in real world conditions if such models are to be reliably integrated with computer vision products for plant disease phenotyping. We train a CNN object detection model to identify foliar symptoms of diseases in cassava (Manihot esculenta Crantz). We then deploy the model in a mobile app and test its performance on mobile images and video of 720 diseased leaflets in an agricultural field in Tanzania. Within each disease category we test two levels of severity of symptoms-mild and pronounced, to assess the model performance for early detection of symptoms. In both severities we see a decrease in performance for real world images and video as measured with the F-1 score. The F-1 score dropped by 32% for pronounced symptoms in real world images (the closest data to the training data) due to a decrease in model recall. If the potential of mobile CNN models are to be realized our data suggest it is crucial to consider tuning recall in order to achieve the desired performance in real world settings. In addition, the varied performance related to different input data (image or video) is an important consideration for design in real world applications.
format Journal Article
id CGSpace105535
institution CGIAR Consortium
language Inglés
publishDate 2019
publishDateRange 2019
publishDateSort 2019
publisher Frontiers Media
publisherStr Frontiers Media
record_format dspace
spelling CGSpace1055352025-11-11T10:03:48Z A mobile-based deep learning model for cassava disease diagnosis Ramcharan, A. McCloskey, P. Baranowski, K. Mbilinyi, N. Mrisho, L. Ndalahwa, M. Legg, James P. Hughes, D.P. cassava diseases plant diseases diagnosis plant condition tanzania Convolutional neural network (CNN) models have the potential to improve plant disease phenotyping where the standard approach is visual diagnostics requiring specialized training. In scenarios where a CNN is deployed on mobile devices, models are presented with new challenges due to lighting and orientation. It is essential for model assessment to be conducted in real world conditions if such models are to be reliably integrated with computer vision products for plant disease phenotyping. We train a CNN object detection model to identify foliar symptoms of diseases in cassava (Manihot esculenta Crantz). We then deploy the model in a mobile app and test its performance on mobile images and video of 720 diseased leaflets in an agricultural field in Tanzania. Within each disease category we test two levels of severity of symptoms-mild and pronounced, to assess the model performance for early detection of symptoms. In both severities we see a decrease in performance for real world images and video as measured with the F-1 score. The F-1 score dropped by 32% for pronounced symptoms in real world images (the closest data to the training data) due to a decrease in model recall. If the potential of mobile CNN models are to be realized our data suggest it is crucial to consider tuning recall in order to achieve the desired performance in real world settings. In addition, the varied performance related to different input data (image or video) is an important consideration for design in real world applications. 2019-03-20 2019-10-28T13:26:15Z 2019-10-28T13:26:15Z Journal Article https://hdl.handle.net/10568/105535 en Open Access application/pdf Frontiers Media Ramcharan, A., McCloskey, P., Baranowski, K., Mbilinyi, N., Mrisho, L., Ndalahwa, M., ... & Hughes, D.P. (2019). A mobile-based deep learning model for cassava disease diagnosis. Frontiers in Plant Science, 10, 272.
spellingShingle cassava
diseases
plant diseases
diagnosis
plant condition
tanzania
Ramcharan, A.
McCloskey, P.
Baranowski, K.
Mbilinyi, N.
Mrisho, L.
Ndalahwa, M.
Legg, James P.
Hughes, D.P.
A mobile-based deep learning model for cassava disease diagnosis
title A mobile-based deep learning model for cassava disease diagnosis
title_full A mobile-based deep learning model for cassava disease diagnosis
title_fullStr A mobile-based deep learning model for cassava disease diagnosis
title_full_unstemmed A mobile-based deep learning model for cassava disease diagnosis
title_short A mobile-based deep learning model for cassava disease diagnosis
title_sort mobile based deep learning model for cassava disease diagnosis
topic cassava
diseases
plant diseases
diagnosis
plant condition
tanzania
url https://hdl.handle.net/10568/105535
work_keys_str_mv AT ramcharana amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mccloskeyp amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT baranowskik amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mbilinyin amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mrishol amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT ndalahwam amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT leggjamesp amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT hughesdp amobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT ramcharana mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mccloskeyp mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT baranowskik mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mbilinyin mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT mrishol mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT ndalahwam mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT leggjamesp mobilebaseddeeplearningmodelforcassavadiseasediagnosis
AT hughesdp mobilebaseddeeplearningmodelforcassavadiseasediagnosis