Development and evaluation of a machine vision-based cotton fertilizer applicator

Abstract

Aim of study: To develop and assess a cotton fertilizer applicator integrated with a Machine Vision Based Embedded System (MVES) to achieve precise and site-specific fertilization. Area of study: The investigation was performed in the Indian Institute of Technology, Kharagpur. Material and methods: The MVES included a cotton detection system with a web camera, processor (computer), and python-based algorithm, and a fertilizer metering control unit with a stepper motor, motor driver, power supply, and microcontroller. The python-based algorithm in the computer predicts the presence (or absence) of cotton plants, whenever an input image is received from the camera. Upon cotton detection, it transforms into a Boolean signal sent to the microcontroller via PySerial communication, which instructs the motor to rotate the metering unit. Motor adjusts the speed of metering unit based on machine speed measured through a hall sensor, ensuring site-specific delivery of metered fertilizer A developed lab setup tested the MVES, experimentally examining performance indicators. Main results: The MVES obtained a MAPE of 5.71% & 8.5%, MAD 0.74 g/plant & 1.12 g/plant for urea and DAP (di-ammonium phosphate), respectively. ANOVA revealed no statistically significant effect of forward speed on the discharge fertilizer amount (p>0.05). For urea, discharge rates ranged from 1.03 g/s (at 10 rpm, 25% exposure length of metering unit) to 40.65 g/s (at 100 rpm, 100% exposure). DAP ranged from 1.43 to 47.66 g/s under similar conditions. Research highlights: The delivered application dosage conformed the recommended dosage. The developed MVES was reliable, had a quick response, and worked properly.