Camera physiological measurement is a fast growing field of computer vision. Remote photoplethysmography (rPPG) uses video cameras (imagers) to measure the peripheral blood volume pulse (BVP). Simply, this enables heart rate measurement via webcams, smartphone cameras and many other imaging devices. The current state-of-the-art methods are supervised deep neural architectures that have large numbers of parameters and a signal number of hyperparameters. Replication of results and benchmarking of new models is critical for scientific progress. However, as with many other applications of deep learning, reliable codebases are not easy to find. We present a comprehensive toolbox, rPPG-Toolbox, containing code for training and evaluating unsupervised and supervised rPPG models: https://github.com/ubicomplab/rPPG-Toolbox