Image processing based Smart Weed Removal and Organic Fertilizer Sprinkling Bot – A Systematic Review

摘要

Most of our country's economy depends on agriculture production. The production of the crop decreases due to unwanted plants (weeds) and the pests' infection around the crop. Weeds are removed manually and also controlled by spraying herbicides. To eradicate the pests, fertilizers are used. Due to the fertilizers and herbicides, the fert ility of the soil decreases and also causes pollution. Manually, these processes are time-taking and need more money to complete both the tasks. In this paper, the weed and diseased plant control using machine learning algorithms based on image processing is discussed. Dataset consists of samples of some healthy and diseased crops. The images in the real field are captured through a vision sensor and pre-processed. The resultant image is compared with the threshold value of the trained dataset. Feature Extraction is done through color, size, and shape. Based on the error rate, the system classifies the plant either as a weed or a diseased plant. If the error rate is high, the plant is classified as a weed and the signal is sent to the robotic arm to pluck the plant through serial communication. If the error rate is less, then it is detected as a diseased plant, and the signal is sent to the sprinkler to spray the organic fertilizer to the infected area. If there is no error, then the plant is treated as a healthy plant. Thus, the machine controls the weeds and the diseased plants around the field autonomously.