Recognition of Homologative Tags for the Automotive Industry via Computer Vision

In recent years, a great change in the automotive industry is the result of innovations introduced by the so-called Industry 4.0, integrating the concept of ”Internet of Things”, Automation and Robotics in plants and manufacturing field. The project presented in this thesis relies on those innovation concepts and illustrates the implementation of an Image Recognition Software application, directed to an automotive big company. The project goal is to help supply chain Operators to perform an effective and efficient check of the homologation tags present on each vehicle. The user can take a picture of the tag to inspect and the application will automatically respond, exploiting Amazon Web Services, returning the result of the implemented controls. These controls are about the correctness, positioning, and integrity of tags. To implement such an application, we used Amazon Web Services (AWS) for computer vision purposes, based on Convolutional Neural Networks to perform text and object detection, and PTC ThingWorx to build Mashups and manipulate data.