Analysis and assessment of human movement are crucial for enhancing athletic performance, preventing injuries, and aiding rehabilitation. This paper applies the most recent iteration of the You Only Look Once (YOLO) family of algorithms (version 8) to resistance training, segmenting humans and gym equipment in real time. Velocity metrics are calculated directly from segmentation masks, adding to the growing body of research concerned with velocity-based training (VBT), an emerging alternative to traditional percentage-based training regimes. Detection performance is assessed in terms of mask mean average precision (mAP50:5:95), and MATLAB ® is used to calculate two important velocity metrics: Peak velocity (PV) and mean velocity (MV). Our trained model achieves 71.7%mAP50:5:95,87.7%mAP50 and 85% F1 score at 0.492 confidence threshold, with an inference speed of 45 fps. Range of motion (RoM), MV, and PV for squat, bench, and deadlift were successfully extracted, demonstrating the potential of pure computer vision methods without requiring costly or inaccessible sensor hardware as in traditional approaches. Our work provides a proof-of-concept for applications of object detection within professional and recreational sports, as well as in rehabilitation settings.