This study presents the development of a monitoring system for adductor canal block procedures, designed to enhance safety and efficiency in medical practice. The system employs landmark detection techniques integrating MediaPipe and YOLO, along with deep learning models to ascertain the patient's posture (prone/prone) and the side being treated (left/right) in real-time. The evaluation conducted under normal conditions demonstrated high accuracy within acceptable limits, with a maximum false warning rate (FPR) of 2.71% and a maximum false detection failure rate (FNR) of 3.31%. Moreover, the system's resilience was validated under abnormal conditions, encompassing 12 failure modes, through the detection of numerous anomalies. The intuitive graphical user interface (GUI) design and real-time warning function facilitated a rapid response environment for healthcare professionals. In the future, the system aims to enhance its responsiveness to real-world medical conditions through the introduction of landmark completion algorithms and the expansion of diverse data sets. The study provided significant results that demonstrate new possibilities in clinical monitoring using AI technology.