This article discusses a novel system for automatically detecting cattle lameness using deep learning-based gait reconstruction and anomaly detection. The system prioritizes a natural approach to minimize stress on cattle and utilizes popular algorithms such as Mask R-CNN and YOLOv8 for cattle detection. The proposed methodology employs a three-point back curvature approach to calculate lameness levels.