Human gait analysis is essential for diagnosing, monitoring, treating, and rehabilitating individuals. To comprehensively evaluate gait, various parameters such as joint angles, ground reaction forces, motion, momentum, velocity, step width, and step length must be measured. This study proposes a novel approach specifically focused on measuring knee angles. Unlike traditional methods that rely on markers and specialized equipment, our approach uses computer vision based on MediaPipe’s posture estimation methodology. Videos for this analysis can be recorded using smartphones or other readily available devices. The results demonstrate that knee angles can be accurately assessed with this markerless method, achieving a mean absolute error of 5.67 \(^{\circ }\) when compared to reference data obtained from VICON. This finding underscores the potential of our approach as a valuable tool for professionals in biomechanics, sports medicine, physiotherapy, and other fields where precise knee angle measurement is crucial for gait analysis.