In recent years, intelligent package registration has been an interesting research topic for different industry areas. In particular, automatic registration based on computer vision methods and product prediction of future requirements are still open research problems. In this work, we developed an intelligent package registration system based on a conveyor belt and a computer vision system able to detect and decode Quick code (QR) information. The computer vision system can detect and decode 6 different QR codes for medicine categories (Loratadine, Ibuprofen, Paracetamol, Omeprazole, Hedillar, and Naproxen). Once the QR code is decoded, the obtained data is uploaded to a data storage registration system in Google Cloud. We conducted experiments to acquire medicine category information for 30 days in a controller environment, and such information was used to train and validate two different supervised machine-learning regression models (linear and polynomial). The proposed models can predict future medical requirements for each category, achieving a Root Mean Square Error (RMSE) of 8.5 for linear regression, and 7.5 for polynomial regression considering the sales for each category. We demonstrated that polynomial regression outperforms linear regression reducing prediction errors up to 15.65%. The system proposed in this work demonstrates the potential of computer vision and machine learning techniques for possible improvements in package registration for pharmaceutical purposes.