Weed Identification and Categorization in Agriculture using Hybrid Deep Learning Algorithm

The use of insecticides, herbicides, and fertilizer in conventional agricultural crop spraying techniques can be decreased thanks to the smart agricultural robotic system. Spraying agro-chemicals to manage weeds and get better crop output is insufficient to meet the world population's food needs. As a result, a elegant and intelligent agricultural farming system is implemented in order to boost crop yield and meet agricultural production targets. Using a mix of Convolutional Neural Networks (CNN) and Long-Short-Term Memory (LSTM), Deep Learning (DL) based algorithms are used to identify and classify weed plants in this paper (LSTM). Convolutional Neural Networks (CNN) has a distinct structure for extracting differential features from input plant images, while LSTM imply granting for combined classification optimization. In order to the proposed method is used to classify nine types of weeds, including prickly weeds, vine weeds, persistent creeping weeds, and three-leaf weeds. We conducted several comprehensive experiments and attained an average classification accuracy of 99.36 percent. The collected results suggest that the CCN-LSTM combo has much better classification capabilities than other well-known techniques.