Weed detection is an indispensable part of precision agriculture, as weeds can severely affect yield and the mismanagement of available resources. Traditionally, weed detection relies mostly on visual inspection or computationally intensive image processing methods that are susceptible to certain errors. A wavelet transform simultaneously represents spatial and frequency information, enhancing the efficiency of weed detection with the necessary precision. Wavelet transformations effectively extract relevant features at several scales and resolutions, making them particularly suitable for distinguishing crops and weeds under varying lighting conditions and textural variations. The proposed method involves pre-processing through contrast enhancement and noise reduction, succeeded by wavelet decomposition, which recovers high-frequency features that contain a significant amount of information regarding weed structures. A support vector machine (SVM), aided by such wavelet features, identifies and classifies weeds from crop regions quite robustly. The experimental results indicate that the proposed methodology outperforms classical ones in terms of better detection rates and reduced false positives. The proposed method gives 92% accuracy and 87.1% F1-score which is better than the existing methods. The presented research work contributes to a promising step toward intelligent and sustainable agricultural practices.