Automated Quality Control for Automobile Manufacturing Using Deep Learning

Abstract

Quality inspection is a critical process in manufacturing industries to ensure product reliability and customer satisfaction. Traditional manual inspection methods are time-consuming, inconsistent, and prone to human error. This paper presents VisionQC, an automated quality control system that leverages deep learning techniques to detect surface defects and damage in automobile components. The system employs a MobileNetV2 convolutional neural network fine-tuned via transfer learning on the Car Damage Detection dataset (Kaggle, ~150 MB), classifying input images as either Damaged / Defect Detected or Good / No Damage using binary sigmoid output. Data augmentation strategies including rotation, zooming, shearing, and horizontal flipping are applied to improve generalisation. The proposed system achieves a validation accuracy of approximately 96.8%, significantly outperforming manual inspection in speed, consistency, and scalability. The results demonstrate the viability of deploying AI-powered vision systems for real-time automobile quality assurance on production lines.