Towards low-cost image-based plant phenotyping using reduced-parameter CNN

Segmentation is the core of most plant phenotyping applications. Current state-of-the-art plant phenotyping applications rely on deep Convolutional Neural Networks (CNNs). However, these networks have many layers and parameters, increasing training and test times. Phenotyping applications relying on...

Full description

Bibliographic Details
Main Authors: Atanbori, John, Chen, Feng, French, Andrew P., Pridmore, Tony P.
Format: Conference or Workshop Item
Language:English
Published: 2018
Online Access:https://eprints.nottingham.ac.uk/54696/
_version_ 1848799066740228096
author Atanbori, John
Chen, Feng
French, Andrew P.
Pridmore, Tony P.
author_facet Atanbori, John
Chen, Feng
French, Andrew P.
Pridmore, Tony P.
author_sort Atanbori, John
building Nottingham Research Data Repository
collection Online Access
description Segmentation is the core of most plant phenotyping applications. Current state-of-the-art plant phenotyping applications rely on deep Convolutional Neural Networks (CNNs). However, these networks have many layers and parameters, increasing training and test times. Phenotyping applications relying on these deep CNNs are also often difficult if not impossible to deploy on limited-resource devices. We present our work which investigates parameter reduction in deep neural networks, a first step to moving plant phenotyping applications in-field and on low-cost devices with limited resources. We re-architect four baseline deep neural networks (creating what we term "Lite CNNs") by reducing their parameters whilst making them deeper to avoid the problem of overfitting. We achieve state-of-the-art, comparable performance on our "Lite" CNNs versus the baselines. We also introduce a simple global hyper-parameter (alpha) that provides an efficient trade-off between parameter-size and accuracy.
first_indexed 2025-11-14T20:29:46Z
format Conference or Workshop Item
id nottingham-54696
institution University of Nottingham Malaysia Campus
institution_category Local University
language English
last_indexed 2025-11-14T20:29:46Z
publishDate 2018
recordtype eprints
repository_type Digital Repository
spelling nottingham-546962018-09-13T07:46:27Z https://eprints.nottingham.ac.uk/54696/ Towards low-cost image-based plant phenotyping using reduced-parameter CNN Atanbori, John Chen, Feng French, Andrew P. Pridmore, Tony P. Segmentation is the core of most plant phenotyping applications. Current state-of-the-art plant phenotyping applications rely on deep Convolutional Neural Networks (CNNs). However, these networks have many layers and parameters, increasing training and test times. Phenotyping applications relying on these deep CNNs are also often difficult if not impossible to deploy on limited-resource devices. We present our work which investigates parameter reduction in deep neural networks, a first step to moving plant phenotyping applications in-field and on low-cost devices with limited resources. We re-architect four baseline deep neural networks (creating what we term "Lite CNNs") by reducing their parameters whilst making them deeper to avoid the problem of overfitting. We achieve state-of-the-art, comparable performance on our "Lite" CNNs versus the baselines. We also introduce a simple global hyper-parameter (alpha) that provides an efficient trade-off between parameter-size and accuracy. 2018-09-06 Conference or Workshop Item PeerReviewed application/pdf en https://eprints.nottingham.ac.uk/54696/1/CNN%200023.pdf Atanbori, John, Chen, Feng, French, Andrew P. and Pridmore, Tony P. (2018) Towards low-cost image-based plant phenotyping using reduced-parameter CNN. In: CVPPP 2018: Workshop on Computer Vision Problems in Plant Phenotyping, 6 Sept 2018, Newcastle upon Tyne, UK.
spellingShingle Atanbori, John
Chen, Feng
French, Andrew P.
Pridmore, Tony P.
Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title_full Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title_fullStr Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title_full_unstemmed Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title_short Towards low-cost image-based plant phenotyping using reduced-parameter CNN
title_sort towards low-cost image-based plant phenotyping using reduced-parameter cnn
url https://eprints.nottingham.ac.uk/54696/