Machine learning integrated into agriculture has really changed the conventional mode of farming by promoting enhancement in the fields of productivity, accuracy, and sustainability. This chapter explores some of the applications of ML in contemporary agriculture, including important techniques such as supervised and unsupervised learning, deep learning, and ensemble methods. Specifically, it assesses five key areas: crop yield prediction, disease detection, soil analysis, weather forecasting, and weed identification. Some of the state-of-the-art ML architectures, like CNNs, RNNs, LSTM, SVM, and more advanced models such as YOLOv7 and SwinRDM, have succeeded in realizing striking improvements in automation and decision-making tasks in agriculture. Recent research within 2020–2025 presents very impressive performance: the ML models’ accuracy in crop yield estimation is as high as 97.6% by ACNN-OBDLSTM, while disease detection can be done with as high as 99% accuracy using SVM and weed detection through YOLOv7. However, there are some crucial challenges that persist related to insufficient data, poor infrastructure, and the generalization performance of models, especially in small-scale farming. It searches for possible solutions through adaptive ML models, integrations of IoT, and scalable methods to make things more accessible. Addressing such challenges, ML has the potential to revolutionize agriculture across the world toward sustainable food production and judicious use of resources. The present study gives an in-depth analysis of the various ML-based innovations in agriculture and offers insight into current achievements and possible future directions to advance the precision of farming with smart systems.