PharmaCV: A Cloud Vision Approach on Generating Labels of Pharmaceutical Drugs

摘要

This paper presents a novel and straightforward way of generating labels from images of common pharmaceutical drugs found in the Philippine market using Cloud Vision. The images used were given labels that indicate the name of the drug and its language code used, the application of the drug, other alternate names of the drug and related labels with their corresponding confidence scores. In addition to that, the uniform resource locator or URL that direct to pages with full and partial matching images were given. Furthermore, the images were stored in a cloud storage bucket which offers good documentation, a high durability of 99.99%, in which the data survives even in the event of simultaneous losses, and easy integration with state-of-the-art Google cloud services such as AutoML, Kubernetes, Compute Engine. The Cloud Vision API was used due to its superb features compared to the similar computer vision APIs that are existing today. This study will be useful to both medical practitioners and drug users since it will be easier for them to gather all the necessary information about a particular drug from its image alone. Furthermore, the system developed in this study can be used in drug dispensing machines where the drug users can be provided enough information about the drug and its applications despite having little knowledge about them.