Welcome to the repository for my Bachelor's thesis, titled 'Quality Control in a Production Line with Deep Learning Supervision'. This project is a testament to the power of deep learning in the realm of quality control, specifically in the context of a production line for syringes. The goal was to develop a system capable of identifying errors in the numbers printed on each syringe, ensuring the highest standards of quality and accuracy
The project utilizes cutting-edge techniques in computer vision, along with a combination of LSTM and self-attention mechanisms, to detect and rectify errors. LSTM, or Long Short-Term Memory, is a type of recurrent neural network that excels at processing sequential data, while self-attention mechanisms allow the model to weigh the importance of different inputs dynamically. Together, these techniques form the backbone of our error detection system.
The code for this project is housed in two main files: models.py, which contains the agents and models, and main.py, which provides a simple framework for implementing the system. The project also leverages the PiCamera library to use a Raspberry Pi 4 as the core engine of the robot, which serves as a separator for quality control. This project is not just a culmination of my Bachelor's thesis, but also a ready-to-use, easy-to-implement system for anyone interested in the intersection of deep learning and quality control. So, dive in, explore the code, and see the power of deep learning in action!