-
Convolutional Neural Network Ensemble Learning for Hyperspectral Imaging-based Blackberry Fruit Ripeness Detection in Uncontrolled Farm Environment
Authors:
Chollette C. Olisah,
Ben Trewhella,
Bo Li,
Melvyn L. Smith,
Benjamin Winstone,
E. Charles Whitfield,
Felicidad Fernández Fernández,
Harriet Duncalfe
Abstract:
Fruit ripeness estimation models have for decades depended on spectral index features or colour-based features, such as mean, standard deviation, skewness, colour moments, and/or histograms for learning traits of fruit ripeness. Recently, few studies have explored the use of deep learning techniques to extract features from images of fruits with visible ripeness cues. However, the blackberry (Rubu…
▽ More
Fruit ripeness estimation models have for decades depended on spectral index features or colour-based features, such as mean, standard deviation, skewness, colour moments, and/or histograms for learning traits of fruit ripeness. Recently, few studies have explored the use of deep learning techniques to extract features from images of fruits with visible ripeness cues. However, the blackberry (Rubus fruticosus) fruit does not show obvious and reliable visible traits of ripeness when mature and therefore poses great difficulty to fruit pickers. The mature blackberry, to the human eye, is black before, during, and post-ripening. To address this engineering application challenge, this paper proposes a novel multi-input convolutional neural network (CNN) ensemble classifier for detecting subtle traits of ripeness in blackberry fruits. The multi-input CNN was created from a pre-trained visual geometry group 16-layer deep convolutional network (VGG16) model trained on the ImageNet dataset. The fully connected layers were optimized for learning traits of ripeness of mature blackberry fruits. The resulting model served as the base for building homogeneous ensemble learners that were ensemble using the stack generalization ensemble (SGE) framework. The input to the network is images acquired with a stereo sensor using visible and near-infrared (VIS-NIR) spectral filters at wavelengths of 700 nm and 770 nm. Through experiments, the proposed model achieved 95.1% accuracy on unseen sets and 90.2% accuracy with in-field conditions. Further experiments reveal that machine sensory is highly and positively correlated to human sensory over blackberry fruit skin texture.
△ Less
Submitted 9 January, 2024;
originally announced January 2024.
-
Global Tweet Mentions of COVID-19
Authors:
Guangqing Chi,
Junjun Yin,
M. Luke Smith,
Yosef Bodovski
Abstract:
Background. After a year and half and over 4 million deaths, the COVID-19 pandemic continues to be widespread, and its related topics continue to dominate the global media. Although COVID-19 diagnoses have been well monitored, neither the impacts of the disease on human behavior and social dynamics nor the effectiveness of policy interventions aimed at its containment are fully understood. Monitor…
▽ More
Background. After a year and half and over 4 million deaths, the COVID-19 pandemic continues to be widespread, and its related topics continue to dominate the global media. Although COVID-19 diagnoses have been well monitored, neither the impacts of the disease on human behavior and social dynamics nor the effectiveness of policy interventions aimed at its containment are fully understood. Monitoring the spatial and temporal patterns of behavior, social dynamics and policy - and then their interrelations - can provide critical information for preparatory action and effective response. Methods. Here we present an open-source dataset of 1.92 million keyword-selected Twitter posts, updated weekly from January 2020 to present, along with a dynamic dashboard showing totals at national and subnational administrative divisions. Results. The dashboard presents 100% of the geotagged tweets that contain keywords or hashtags related COVID-19. We validated our inclusion criteria using a machine learning-based text classifier and found that 88% of the selected tweets were correctly labeled as related to COVID-19. With this information we tested the correlation between tweets and covid diagnosis from January 1, 2020 through December 31, 2020 and see a decreasing correlation across time. Conclusions. With emerging COVID variants but ongoing vaccine hesitancy and resistance, this dataset could be used by researchers to study numerous aspects of COVID-19 and provide valuable insights for preparing future pandemics.
△ Less
Submitted 17 August, 2021; v1 submitted 13 August, 2021;
originally announced August 2021.
-
Concurrent Composition and Algebras of Events, Actions, and Processes
Authors:
Mark Burgin a,
Marc L. Smith
Abstract:
There are many different models of concurrent processes. The goal of this work is to introduce a common formalized framework for current research in this area and to eliminate shortcomings of existing models of concurrency. Following up the previous research of the authors and other researchers on concurrency, here we build a high-level metamodel EAP (event-action-process) for concurrent process…
▽ More
There are many different models of concurrent processes. The goal of this work is to introduce a common formalized framework for current research in this area and to eliminate shortcomings of existing models of concurrency. Following up the previous research of the authors and other researchers on concurrency, here we build a high-level metamodel EAP (event-action-process) for concurrent processes. This metamodel comprises a variety of other models of concurrent processes. We shape mathematical models for, and study events, actions, and processes in relation to important practical problems, such as communication in networks, concurrent programming, and distributed computations. In the third section of the work, a three-level algebra of events, actions and processes is constructed and studied as a new stage of algebra for concurrent processes. Relations between EAP process algebra and other models of concurrency are considered in the fourth section of this work.
△ Less
Submitted 20 March, 2008;
originally announced March 2008.
