Agriculture and weed management are intrinsically linked, as the presence of weeds can significantly impact crop yields and the overall health of agricultural systems. Convolutional Neural Networks (CNNs) have shown considerable promise in automating weed detection using plant seedling datasets. The proposed CNN model is validated with various optimizers to enhance the performance of weed detection. This study evaluates the impact of different optimizers, Stochastic Gradient Descent (SGD) and Root Mean Square Propagation (RMSprop), on the proposed CNN model. Through experimentation, we found that the choice of optimizer significantly affects detection accuracy. Our proposed model is efficient, performing better with the appropriate selection of optimizers and demonstrating its adaptability to different optimization techniques. The performance of the proposed model is evaluated by determining the accuracy, recall, precision, F1 score, confusion matrix, macro average precision, macro average recall, weighted average. Among the tested optimizers, RMSprop demonstrated superior performance, achieving an accuracy of 97.38%, validating the robustness and effectiveness of our proposed model.