Chu, Duc Ha and Nguyen, Quoc Trung and Tong, Van Hai and Ta, Hong Linh and La, Viet Hong and Vuong, Quang Huy and Vu, Minh Trung and Pham, Minh Trien
ESTABLISHMENT OF THE DIGITAL TOOLS FOR PRECISION AGRICULTURE BY MACHINE LEARNING.
In: Agricultural Biotechnology: Challenges and Opportunities.
Abstract
Precision agriculture has been considered as one of the key components of the digital
transformation in Vietnam. In the view of information and communications technology, five major
cases, including smart-crop monitoring, drone farming, smart-livestock monitoring, autonomous
farming machinery, and smart-building and -equipment management have the potential to radically
transform many aspects of agriculture. However, the establishment of digital tools used in smart
agriculture programs has been still lacking. Of our interest, we reported two out of many cases of
Internet-of-Things- (IoT-) based tools for the research in agriculture. In the first case, we investigated
an electronic trap for automated monitoring of fall armyworm (FAW). Briefly, FAW (Spodoptera
frugiperda) has been reported as one of the most devastating pests that can attack maize (Zea mays) at
all growth stages. Since the first occurrence of FAW in Vietnam has been reported in 2019, this
lepidopteran pest had caused huge damage to maize production in the Northern provinces of Vietnam.
Thus, monitoring, identification, and management of FAW in the fields become one critical task for
sustainable agricultural production. As the result, we introduced an automated FAW (adult moths)
counting system based on the traditional pheromone traps. The IoT sensors have been merged into the
instrument to count the frequency of adult moths and together record the temperature and humidity data.
The general data, including the real-time amount of trapped insects and environmental conditions, has
been analyzed based on the machine-learning method, consequently, send to the Internet browser and
applications. In the second case, we constructed a cost-effective phenotyping machine for automated
seed imaging. Particularly, agronomists have an issue with the estimation of various typical
characteristics of seeds, like length, width, mass, the color of the skin, and pubescence. By using the
computer vision approach, we generated an easy-to-use tool for automatically measuring the general
features of crop seeds. The construction of this tool can significantly replace labors with only simple
operations. Taken together, our tools could significantly provide a collection of digital tools for
supporting the digital transformation in research and development in the agriculture sector.
Actions (login required)
 |
View Item |
