IoT-based Sensing System for Thrips Pest and Disease Management in Onion Crop

Onion (Allium cepa L.) is an economically valuable crop. Due to various diseases and pests, such as Thrips, the average onion crop yield decreases by 30 – 50 %. Hence, it is essential to accurately detect diseases and pest in the farm to implement appropriate control measures. Early detection of the disease and possibility of thrip pest attack can control the spread, thereby reducing the yield loss of onion. By measuring field parameters using sensor-based Internet of Thing (IoT) system and machine learning (ML) based analysis, the probability of a thrip attack and its life cycle can be correlated with local weather and soil conditions. In addition, accurate detection of diseases can be achieved by inspection of visual images using image processing and machine learning techniques. Using a combination of IoT-based monitoring system and ML based analysis, this paper detects diseases and investigates the life cycle of thrips/pest. The paper focuses on 1) developing a mechanism to record and correlate various farm parameters (such as ambient temperature, relative humidity (RH), soil moisture and temperature) in real-time to understand the thrip cycle, 2) capturing images for ML-based analysis 3) disseminating disease control techniques. The systematic study shows that three of the 12 pesticide calendar sprays can be avoided by correlation of weather and soil parameters. Further, onion diseases can be detected with 96.3 % accuracy using ML-based image analysis.