In today’s manufacturing industry, there is a growing need for precise and efficient quality control systems. This article proposes a real-time solution for identifying missing or damaged parts in a can by integrating computer vision, deep learning networks, robotics, and artificial intelligence. A collaborative robotic system was used to improve manufacturing efficiency. The system consists of a Doosan robot equipped with a 2.5D vision system and suction end-effectors for accurate object manipulation. The study involved collecting and augmenting a dataset using the Hough Transformation technique to isolate individual cans. Three pre-trained models, namely VGG16, MobileNet, and ResNet101, were used to detect defects on the can top through transfer learning. The performance of the models was evaluated in terms of accuracy, speed, and efficiency. The algorithm was successfully integrated into the robot, enabling it to perform real-time defect detection and classification autonomously. The implementation of this system showcased significant improvements in defect detection, highlighting the potential of automated technologies in industrial applications. The study also explores the implications of these findings for future enhancements, including the potential for increased system precision and the broader application of computer vision and artificial intelligence in manufacturing processes.