Smarter Pest Identification Technology (SPIDTECH): a mobile application for digital identification and remote monitoring of insect pests and diseases of major crops in the Philippines

Issue Date

12-2021

Abstract

Correct identification and proper monitoring are vital components of integrated pest and disease management. The Android application – Smarter Pest Identification Technology (SPIDTECH) – was developed for the digital identification and remote monitoring of insect pests and diseases of rice, corn, coffee, cacao, banana, coconut, sugarcane, tomato, and soybean in the Philippines. SPIDTECH has three main features: the digital identification of insect pests and diseases using a smartphone camera; the digital library that contains images, identification signs, life stages, management practices, and other pertinent information about a pest or disease; and the remote monitoring that enables real-time mapping of reports through user-contributed images and GPS points. The application can identify 71 insects and 63 diseases, while the library features a collection of 104 insects and 89 diseases available in English and Filipino. The identification uses a pretrained MobileNetV2, a convolutional neural network design by Google for mobile devices. The models were retrained using the pest and disease image dataset collected specifically for this study and logging 55.7–73.3% accuracy. If connected to the internet, devices send image data from the field that are validated and used for model retraining. SPIDTECH recorded more than 7,400 device downloads from March 2019–July 2021 and received 4.8 out of 5 user review ratings. Based on the user-provided registration data in the same period, it has more than 5,600 registered users from 81 provinces with an average user acquisition rate of 266 new users per month. With the current features deployed, the application received more than 8,000 identification requests for different crops. The application has gathered a significant number of users necessary to evaluate the feasibility of the application in assisting in pest and disease identification and remote monitoring for the major crops in the Philippines.

Source or Periodical Title

Philippine Journal of Science

ISSN

0031-7683

Volume

150

Issue

6B

Page

1811-1821

Document Type

Article

Frequency

quarterly

Physical Description

illustrations; map

Language

English

En – AGROVOC descriptors

PLANT PROTECTION; PEST MONITORING; CROP MONITORING; MONITORING; INTEGRATED PEST MANAGEMENT; INTEGRATED DISEASE MANAGEMENT; MACHINE LEARNING; IDENTIFICATION; PHILIPPINES

Share

COinS