In this paper, an unobtrusive, fully-automated,model-based algorithm is proposed, which is able to estimate thepose of a cyclist and track it over time to extract importantpose and movement parameters. Applied techniques includebackground subtraction, skin detection, principal componentanalysis and template matching. The proposed algorithm is robustagainst variations in human appearance such as body size andclothing. Besides pose estimation, we propose a non-contact,vision-based heart rate estimation algorithm. The algorithm isable to detect and track the subtle temporal skin color changescaused by the flowing blood through the vessels and extractthe corresponding cardiovascular Blood Volume Pulse (BVP)signal. Our proposed heart rate estimation algorithm includes(fore-)head tracking, independent component analysis and motionfiltering, to eliminate frequencies caused by human motion. Ourpose estimation and heart rate detection systems are successfullyvalidated on manually generated ground-truth data. The meanabsolute body part orientation error is between 0.8 and 7.4 degrees for the pose estimation algorithm and between 1.9 and3.9 beats per minute for the heart rate detection algorithm. By combining both systems, fully-automatic, non-contact cyclingperformance analysis can be performed based on video input only.