A key challenge in Industry 4.0 is integrating advanced technologies to enhance overall system efficiency. While collaborative robots (cobots) and deep learning-based object detection models have advanced, their deployment for vision-based tasks with robotic arms remains understudied. In this research, a vision-set mounted on a robotic arm is tested for sorting the mechanical fasteners. Three object detection models i.e., YOLO, SSD, and Faster R-CNN have been trained on over 2500 images and their sorting performance is evaluated for static and real-time object detection using vision-set. The trained models were validated through deployment with robotic arm. YOLO has proven to be the most effective algorithm considering training, speed and accuracy while the other models lacked in certain aspects one way or the other.