Subjective visual assessment of gait analysis is prone to observer bias and variability among evaluators. This study trained and validated a deep learning–based markerless gait analysis model for use in dogs.