Urban forestry is of pivotal significance in the context of fostering sustainable and resilient cities. However, conventional monitoring and management methodologies are characterized by their labor-intensiveness and inefficiency. Recent advancements in machine learning (ML) offer transformative opportunities to enhance the automation, scalability, and accuracy of urban forest analysis. This critical review discusses the integration of ML with advanced remote sensing technologies—including satellite imagery, LiDAR, photogrammetry, and mobile mapping—to revolutionize urban forestry practices. In comparison to previous studies that primarily focus on isolated applications of ML, this review provides a comprehensive synthesis of state-of-the-art methodologies, bridging the gap between ML-driven automation and practical urban forestry management. Key topics include vegetation classification, point cloud data extraction, disease detection and species distribution mapping. Beyond these fundamental tasks, the study highlights pioneering applications such as the creation of digital twins of urban forests, which enable real-time monitoring and predictive modeling of tree health, distribution, and ecosystem services. By critically evaluating existing methodologies, their effectiveness, and emerging trends, this paper identifies the most promising ML strategies for optimizing urban forestry management. Furthermore, this review outlines current challenges, such as data availability, algorithmic biases, and computational constraints, while proposing future research directions to enhance the integration of ML in urban green space planning. This study presents a structured assessment of ML applications in urban forestry and serves as a valuable reference for researchers, policy makers and urban planners. The assessments promote the effective use of ML to enhance the ecological, social and economic functions of urban forests, supporting the long-term health and sustainability of these essential ecosystems.