Motion capture system is used to measure 3-dimensional kinematics of rigid bodies. However, the kinematics measurement of human motion using a marker-based motion capture system is available only in a lab space where multiple cameras are installed. In addition, the marker-based motion capture system requires a long preparation time to attach active or passive markers onto anatomical landmarks. Although markerless motion capture system or IMU-based motion capture system have been developed to overcome the disadvantages of the marker-based motion capture system, those systems still have many aspects to improve. In this study, we developed a portable marker-based motion capture system, which does not require the installation of multiple cameras prior to measurement and thus allows us to conveniently measure human body kinematics. The system consists of a Raspberry Pi 3B, two Raspberry Pi camera modules, and LED-based marker sets. The parameters for the stereo vision system include a resolution of 640x480, a frame rate of 90 fps, and an ISO value of 600. The calibration was performed using a LED checkerboard. Then, the LED marker sets were placed on each rigid body. The results show that the error in the measurement was less than 1.2% or 0.3 mm.