Implementation of Multi-feature Image Identification Using Supervised Binary Classification and Maximum Likelihood Estimate in a Robotic Herbicide Spray Machine

• Author(s): Dr. Mohammed Aliyu Gadam ; Abubakar AbdulKadir
• Paper ID: 1705587
• Page: 131-139
• Published Date: 18-03-2024
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 7 Issue 9 March-2024

Abstract

Agriculture sector is adopting new technologies, which seems very promising as it will enable farm productivity and profitability. Weed control is one of Agriculture's few operations that is still not yet fully mechanized, labour-intensive and holds potential for automation. The great demand for labour as a major problem in weed management paved the way for the farmers to look for alternative through the spraying of harmful pesticides. However, this led to several health problems for the farmers and the people who are consuming it. Under natural growing conditions, weeds are generally distributed in small patches, but farmers often uniformly spray herbicide in their fields, which is not in agreement with sustainable agriculture development and increases the cost of crop production. Autonomous robotic weed control systems with multi-feature image identification using Convolution Neural Network (CNN) supervised binary classification and maximum likelihood estimate (MLE) have proven to be a viable alternative, leveraging on the powerful representation learning capabilities of CNNs while incorporating the probabilistic modeling and parameter estimation of logistic regression. The experimental result of the proposed system has achieved an average of 97% classification rate for weed and plants in real time when implemented in small machines. The system hold promise to potentially improve agriculture's sustainability, lessen its negative effects on the environment, and lower the overall cost of production while lowering its current reliance on pesticides.

Keywords

Artificial Intelligence, Autonomous Machines, Binary Classification, Convolution Neural Network (CNN), Weed Control

Citations

IRE Journals:
Dr. Mohammed Aliyu Gadam , Abubakar AbdulKadir "Implementation of Multi-feature Image Identification Using Supervised Binary Classification and Maximum Likelihood Estimate in a Robotic Herbicide Spray Machine" Iconic Research And Engineering Journals Volume 7 Issue 9 2024 Page 131-139

IEEE:
Dr. Mohammed Aliyu Gadam , Abubakar AbdulKadir "Implementation of Multi-feature Image Identification Using Supervised Binary Classification and Maximum Likelihood Estimate in a Robotic Herbicide Spray Machine" Iconic Research And Engineering Journals, 7(9)