Publication

Abstract : Weed management in agricultural farms is one of the most crucial activities that the farmers should do to get a good yield of crops. This paper covers the implementation of a model robot that can detect the weeds inside farmland and display the coordinates of the weeds; this can be achieved by introducing a machine learning technique to the system for categorizing the weeds and crops from farmland. The system utilizes the TensorFlow library to autonomously learn distinctive features of weeds and crops, enabling it to differentiate between the two categories for accurate detection. To achieve this, the system was trained using a carefully created dataset, allowing the system to perceive the crucial features associated with weeds and crops. Once the system identifies a weed, it can autonomously adjust the PWM (Pulse Width Modulation) to drive the robot's motors to move towards the weed. Furthermore, it can independently determine and fetch the precise central coordinates of the identified weed.