Classification of metal screw defect detection using FOMO on edge impulse / Muhammad Imran Daing

Surface defects in metal screws are typically identified through manual inspection, which can be prone to human error. The introduction of deep learning, particularly in visual detection, offers a significant improvement in the effectiveness and precision of defect identification. This project uses the FOMO (Faster Objects, More Objects) algorithm to detect surface flaws on metal screws. FOMO is optimized for real-time applications, making it suitable for edge devices with limited resources such as microcontrollers. The model processes images to identify defects by analyzing features such as shape, texture, and structural integrity. By enabling defect detection directly on compact, low-power devices, this method reduces the need for expensive hardware and complex setups. The results demonstrate that FOMO not only achieves high accuracy of 94.5% in defect detection but also operates efficiently, with minimal latency, making it ideal for real-time applications. Furthermore, this approach underscores the growing role of deep learning in automating quality control processes, offering a fast, reliable, and scalable solution for the manufacturing industry.