Deep Learning Transforming the Manufacturing Industry: A Case Study

Abstract

The traditional manufacturing industry uses decades-old technology and manual processes that are expensive, laborious, redundant, and a slight human error can cost millions of dollars. With the industrial revolution 4.0, industries are adopting intelligent additions and modifications into their systems involving robotics and internet of things (IoT) based technology, which can help in their operations. Artificial intelligence and machine learning have revolutionized every field, and the manufacturing industry has also started to reap its benefits. In this case study, we have explained the applications of deep learning-based approaches contributing to the automation of surveillance of production lines. Several cameras are installed to collect live feeds on the floor lines where the manually operated cranes and different tasks it performs are detected and tracked by state-of-the-art Convolutional Neural Network-based YOLOv3, an object detection model, and DeepSORT tracking algorithm. The post-processing is done on the matrices generated. It varies based on the use case and helps the manufacturers with automated inspection, quality control, and labor productivity. This process will save millions of dollars per year just in the crane maintenance operation for our industry partner's client. The developed technology can be replicated in almost every manufacturing industry using traditional methods and provide them better key insights into their production lines.