Weed control is an important issue in modern agriculture, which heavily impacts crop yield and farm productivity. Manual weeding is quite labor-intensive and time-consuming, while excessive use of herbicides is harmful to the environment and soil. This paper introduces a novel deep learning-based automated weed detection system that uses some advanced image-processing techniques like grayscale conversion, edge detection, and segmentation to precisely identify weeds in agricultural fields. Using a Convolutional Neural Network (CNN), the system achieves an accuracy rate of more than 98% in distinguishing crops from weeds based on features such as shape, texture, and regions of interest extracted from 1305 field images. This approach reduces herbicide dependency and labor costs, promotes environmental sustainability, and helps to improve crop health. The key findings of this work open doors to integrating autonomous weed removal mechanisms into the future, thus paving a way towards fully autonomous management of weeds.