Recognition of Medicine Using CNN for Visually Impaired

Abstract

The capacity to carry out routine activities, such as taking medication, is impaired by low vision. While those with normal vision may simply use their eyes, the blind need supplementary aid in order to find and recognize medications. Therefore, it is vital to get timely assistance in order to prevent drug abuse. Thirty people who are either legally blind or legally deaf were interviewed about the medications they use at one of three support centers. The participants' present techniques for locating their medicine have limitations, which has led to their taking inadvertent irregular amounts. We built an Android-based drug detection model and braille embosser system in response to the information gleaned from the interview. The CNN-based recognition model achieves 99.6% accuracy when presented with an image of a medication obtained using the device's built-in camera. Additionally, the categorization findings may be printed as a braille label for future identification without a smartphone using a low-cost braille embosser that connects to a smartphone through Bluetooth.