This study presents the design and implementation of a low-cost, markerless body tracking prototype for gait analysis using computer vision techniques. The system utilizes MediaPipe Holistic and the Python programming language to detect and track human posture in real-time, enabling the estimation of joint angles at the elbows, hips, knees, and ankles throughout the gait cycle. A user-friendly graphical interface was developed using Tkinter, allowing non-specialized operators to perform assessments efficiently. The prototype was tested in a controlled environment consisting of two cameras for lateral and posterior views, as well as a defined walking path. Additionally, the prototype was evaluated by a young child from the Special Education Carlos Garbay School in Riobamba, Ecuador, demonstrating effective joint tracking and gait cycle analysis over an 8-second sequence. Graphical outputs illustrated joint flexion and extension patterns, aligning with standard biomechanical models. The findings suggest the feasibility of deploying this accessible and affordable tool in settings with limited resources, highlighting its relevance for telerehabilitation and early diagnosis of neuromuscular disorders.