EFFECTIVENESS OF IMAGE-BASED PLANT DISEASE DEECTION SYSTEM

Abstract

Image-based Plant Disease Detection Systems (IPDDSs) are increasingly used in agriculture for detecting plant diseases. A good number of models and techniques are used for this purpose. Usually, the effectiveness of these models and techniques is studied from developers’ perspectives. This study intended to find out the effectiveness of IPDDS in detecting plant diseases from the users’ perspectives. A total of 384 IPDDS users across Bangladesh were selected randomly and interviewed during the period from May 01, 2021 to September 30, 2021. Major findings indicated that the majority of the users (77.90%) perceived IPDDS as moderately effective while 19.50% of the users perceived IPDDS as highly effective, and 2.60% perceived it as less effective in detecting plant diseases and getting plant disease-related services. Among the effectiveness dimensions ‘user satisfaction’ ranked first followed by ‘system accuracy’, ‘user friendliness’, ‘content quality’, ‘offline usability’, and ‘device responsiveness’ based on the Average Standardized Effectiveness Index (ASEI). According to the descending order of the Standardized Effectiveness Index (SEI), ‘on-time service’ ranked first followed by ‘time needed for getting service’, ‘service cost’, ‘delivery of the service’, ‘ease of the system’, etc. Step-wise multiple regression showed that out of 14 selected characteristics of the IPDDS users, four (4) characteristics namely, ‘time saved’, ‘knowledge on plant disease management’, ‘benefits obtained by using IPDDS’, and ‘use of ICT’ had significant positive contribution and ‘obstacles faced in using IPDDS’ had significant negative contribution to their perceived effectiveness of IPDDS. These five (5) variables explained 34.5 percent of the total variation in the effectiveness of IPDDS. Each of the Five (5) variables had indirect effect on the effectiveness of IPDDS through the other four (4) variables. According to the descending order of Standard Benefit Index (SBI), ‘reduce time for detecting plant disease’ ranked first followed by ‘increase service accessibility’, ‘increase service quality’, ‘increase respondents’ knowledge’, etc. According to the descending order of Standardized Obstacle Index (SOI) ‘illiteracy’, ranked first followed by ‘inability of disease detection at an early stage’, ‘unavailability of the smart phone’, ‘ICT phobia’, ‘inadequate extension service to support the use of IPDDS’, etc. Frequently suggested suggestions by the users included covering all stages of the disease, providing farmers’ training, introducing off-line system, comparison system, voice interactive system, intelligent system, image uploading option, etc. Based on the suggestions of the users and the findings of the research, a functional model for improving DPPIS is proposed.