Context TruMedicines has trained a deep convolutional neural network to autoencode and retrieve a saved image, from a large image dataset based on the random pattern of dots on the surface of the pharmaceutical tablet (pill). Using a mobile phone app a user can query the image datebase and verify the query pill is not counterfeit and is authentic, additional meta data can be displayed to the user: manf date, manf location, drug expiration date, drug strength, adverse reactions etc. Content TruMedicines Pharmaceutical images of 252 speckled pill images. We have convoluted the images to create 20,000 training database by: rotations, grey scale, black and white, added noise, non-pill images, images are 292px x 292px in jpeg format In this playground competition, Kagglers are challenged to develop deep Convolutional Neural Network and hash codes to accurately identify images of pills and quickly retrieved from our database. Jpeg images of pills can be autoencoded using a CNN and retrieved using a CNN hashing code index. Our Android app takes a phone of a pill and sends a query to the image database for a match, then returns meta data abut the pill: manf date, expiration date, ingredients, adverse reactions etc. Techniques from computer vision alongside other current technologies can make recognition of non-counterfeit, medications cheaper, faster, and more reliable. Acknowledgements Special Thanks to Microsoft Paul Debaun and Steve Borg and NWCadence, Bellevue WA for their assistance Inspiration TruMedicines is using machine learning on a mobile app to stop the spread of counterfeit medicines around the world. Every year the World Health Organization WHO estimates 1 million people die or become disabled due to counterfeit medicine.