Smart Pill Identification System with Medication Advisor

Ms. B Rupa Devi
Associate Professor,Dept of AIDS, AnnamacharyaInstitute of Technology and
Sciences, Tirupati-517501, India.rupadevi.aitt@annamacharyagroup.org
ORCID: 0009-0005-1298-737X
K Divya Sree
UG Student, Dept. Of ArtificialIntelligence and Data science,Annamacharya Institute of
Technology and Sciences,Tirupati, India.kuppamdivyasree@gmail.com

R Ezhil
UG Student, Dept Of ArtificialIntelligence and Data science,Annamacharya Institute of
Technology and Sciences,Tirupati, India.ezhilrajeshkanna41@gmail.com

S Aashifa
UG Student, Dept. Of ArtificialIntelligence and Data science,Annamacharya Institute of
Technology and Sciences,Tirupati, India.
aashifasyed05@gmail.com
V Bala Adithya
UG Student, Dept Of ArtificialIntelligence and Data science,
Annamacharya Institute ofTechnology and Sciences,Tirupati, India.
vacchaadithya@gmail.com

Abstract— In this paper, we outline a Smart Pill Identification System that is based on deep learning techniques, and attempts to improve medication safety by creating an image classification paradigm. The architecture that will be proposed uses a transfer learning Convolutional Neural Network (CNN) thatutilises lightweight MobileNet classify pharmaceutical pills with high precision taking advantage of visual discriminative, visual features, i.e.,the colour, morphology, texture, and imprint pattern. Inaddition to identification, the system providescomprehensive pill-related data, which includes the pharmaceutical name, therapeutic indications, dosage rule, time of administration, potential adverse outcomesand contraindications. The system is created to help both the health professionals and the patients eliminate medication errors that come about due to visual homogeneity of the pills. The choice of MobileNet has been made due to it quality equilibrium between classification performance and balanced computing efficiency and allows application to portable andresource-constrained devices. an interactive medical consultative feature allows the users to ask overall medical questions that are related tomedication with the responses being enhanced by relevant safety warnings. Keywords—Deep Learning, Pill Identification, MobileNet, CNN, Healthcare AI, Medication Safety, Image Classification