Background. The markerless motion capture system (MLS) is a new technology that uses deep learning to detect body segments from digital images. This study tested the agreement between the MLS and a force-plate based system (FPS, “gold standard”) to quantify stability control and motor performance (MP) during gait initiation (GI). Methods. Healthy adults (young and elderly) and patients with Parkinson’s disease (PD) performed GI series at spontaneous and maximal velocity on an FPS while being filmed by an MLS. Signals from both systems were used to compute the peak of forward center-of-mass velocity (indicator of MP) and the braking index (BI, indicator of stability control). Results. Descriptive statistics indicated that both systems detected between-groups differences and velocity effects similarly, while a Bland-Altman (BA) plot anal-ysis showed that BI and MP mean biases were virtually zero in all groups and conditions. Bayes factor 01 indicated strong (for BI) and moderate (for MP) evidence that both systems provided equivalent values. However, trial-by-trial analysis of BA plots revealed the possibility of differ-ences > 10% between the two systems. Conclusion. Although non-negligible differences do occur, MLS appears to be as efficient as FPS in detecting PD and velocity condition effects on BI and MP