Smart farming is the contemporary way of practicing agriculture by leveraging the latest technologies related to AI, computer vision, and automation for better crop production and resource utilization. The proposed research, in this regard, presents an advanced Weed and Crop Classification System that could classify different crop and weed species with high accuracy based on image analysis. The system includes a lightweight yet very high-performance deep convolutional neural model called EfficientNet-B3, ensuring reliable feature extraction and classification accuracy even if the CPU resources are constrained. Adaptive Histogram Equalization with Contrast Limited is applied as an image preprocessing technique to enhance local contrast and make subtle features of the input images more visible before classification. Detailed experimental investigation showed that the proposed model achieved an overall classification accuracy of more than 93% that could reliably differentiate crop and weed species. Such performance indicates the possibility of including AI-driven image analysis into precision agriculture systems to enable data-driven decision-making, reduce human labor, and increase the overall efficiency of weed management practices.