Machine learning enabled identification and real-time prediction of living plants’ stress using terahertz waves

Received: 14 Mar 2022, Revised: 02 Apr 2022, Accepted: 22 July 2022, Available online: 28 Sep 2022, Version of Record: 28 Sep 2022

Adnan Zahid a b, Kia Dashtipour b, Hasan T. Abbas b, Ismail Ben Mabrouk c, Muath Al-Hasan c, Aifeng Ren d, Muhammad A. Imran b, Akram Alomainy e, Qammer H. Abbasi b
a
School of Engineering and Physical Science, Heriot-Watt University, Edinburgh, EH144AS, UK
b
James Watt School of Engineering, University of Glasgow, Glasgow, G128QQ, UK
c
College of Engineering, Al-Ain University, Abu Dhabi, United Arab Emirates
d
School of Electronic Engineering, Xidian University, Xi'an, Shaanxi, China
e
School of Electronic Engineering and Computer Science, Queen Mary University of London, London, UK

Abstract


Considering the ongoing climate transformations, the appropriate and reliable phenotyping information of plant leaves is quite significant for early detection of disease, yield improvement. In real-life digital agricultural environment, the real-time prediction and identification of living plants leaves has immensely grown in recent years. Hence, cost-effective and automated and timely detection of plans species is vital for sustainable agriculture. This paper presents a novel, non-invasive method aiming to establish a feasible, and viable technique for the precise identification and observation of altering behaviour of plants species at cellular level for four consecutive days by integrating machine learning (ML) and THz with a swissto12 materials characterization kit (MCK) in the frequency range of 0.75 to 1.1 THz. For this purpose, measurements observations data of seven various living plants leaves were determined and incorporate three different ML algorithms such as random forest (RF), support vector machine, (SVM), and K-nearest neighbour (KNN). The results demonstrated that RF exhibited higher accuracy of 98.87% followed by KNN and SVM with an accuracy of 94.64% and 89.67%, respectively, for precise detection of different leaves by observing their morphological features. In addition, RF outperformed other classifiers for determination of water-stressed leaves and having an accuracy of 99.42%. It is envisioned that proposed study can be proven beneficial and vital in digital agriculture technology for the timely detection of plants species to significantly help in mitigate yield and economic losses and improve crops quality.

Keywords
Terahertz sensing
Plants health
Machine learning



Description



   

Indexed in scopus

https://www.scopus.com/authid/detail.uri?authorId=57198859755
      

Article metrics

10.31763/DSJ.v5i1.1674 Abstract views : | PDF views :

   

Cite

   

Full Text

Download

Conflict of interest


“Authors state no conflict of interest”


Funding Information


This research received no external funding or grants


Peer review:


Peer review under responsibility of Defence Science Journal


Ethics approval:


Not applicable.


Consent for publication:


Not applicable.


Acknowledgements:


None.