In recent years, there has been a significant increase in industrial production volume, necessitating more dynamic and controlled production processes to ensure agility and reliability. One effective approach to achieving these goals is the use of Machine Learning (ML), which enables machines to learn from data and develop models capable of identifying new information. The Haar-Cascade method, known for its effectiveness in object detection, was utilized to automate the identification of a membrane in automotive taillights. A low-cost prototype was developed, employing computer vision to automatically verify the presence of the membrane. The prototype includes a controlled measurement environment and software developed in Python using the OpenCV library for image analysis. The classifier demonstrated extremely high accuracy, with a 0% error rate in controlled tests. Even under non-controlled conditions, the error rate remained very low. The detection process proved to be instantaneous and stable, even with object movement, highlighting the robust-ness of the Haar-Cascade method and the effectiveness of the proposed solution in enhancing quality control processes.