Performance optimization is a continuous pursuit in athletics, particularly in sports such as javelin throw, where even subtle improvements in technique can lead to significant performance gains. This paper presents a novel approach to analyzing the biomechanics behind the javelin throw with the help of computer vision techniques. Our methodology employs body position tracking with the help of proprietary landmark detection algorithms to analyze the athlete’s throwing technique. The athlete’s body position is accurately tracked during the last seven steps of the javelin throw, called the crossover phase, including the critical release step. Additionally, we utilized landmark detection algorithms to identify several critical anatomical landmarks and tracked eight key body angles essential for the javelin throw technique. Also, we custom-trained our own YOLOv7 model for javelin detection and tracking. We used a dataset that consisted of video footage and images sourced from the internet to train the custom YOLOv7 model. The model allowed accurate tracking of the position and angle of the javelin throughout the approach phase. By precisely monitoring body positions and tracking the path of the javelin in the crossover phase, our methodology provides valuable insights into the biomechanics of this sport.