iSpy – CSE Senior Design

Team Name

We Discover

Timeline

Fall 2018 – Spring 2019

Students

Abdul Mannan
Cameron Howard
Gaganjeet Singh
Ivy Moore
William Truong

Abstract

iSpy is an Android application that uses the camera on the user’s phone to take a photo of an item. That photo is sent to a database where it is then classified. As it recognizes the object using machine learning, it will search online for local stores that carry it. Using the user’s GPS, this application will find nearby stores which still have the product in stock. Once various stores are found, the user will be notified on what stores are closet or offer the cheapest prices to buy their item.

Background

The objective of this project is to provide consumers with a handy way to find products that they need locally. When users search for items online, they are usually lead to an e-commerce site, which means that users typically won’t be able to receive their wanted items in the same day. Instead of waiting, with our app, consumers will be able to find out where the closest store is that carries their wanted item (using their phone’s camera). If a store is a close enough, users will not have to wait for an e-commerce site to ship their item. The goal of our product is to see a future where finding everyday necessities are only a few taps away.

Project Requirements

For customer requirements, they will need to have Camera and Internet availability. We will also want our results list to function correct for the customer to use.

For packaging requirements, as of right now they need Android OS compatibility as well as access to the Google play store to download our application.

For performance requirements, we need fast web service response times, smooth application transitions and for the photo to send quickly as well as the queries to run efficiently.

System Overview

The system has three layers, the client’s side (the phone), the web services, and the server. Because

our application is software heavy, there are not many layers. The system is designed to be simple

but effective. The Client and Server layers both communicate through the Web Services layer, which

contains the HTTP Request and the API Calls components

Results

Currently, the project has matched most of the constraints. We did end up changing our machine learning algorithm due to the previous one having poor accuracy of only 30% and very slow compile times. We were originally using our own programmed algorithm and now are using an already implemented on within Google Cloud Platform to make it easier and quicker to train our data as well as have higher accuracy of classification.

The projects focus on clothing and branching out to other items was also changed. Right now we are picking items that are rather obvious in their classification due to the lack of large amounts of training data.

Overall, the project was successful. We were able to get the camera working, the photo sent to the ML algorithm and it processed with list results in order to buy that item. If we had more well-classified data sets, we could train our algorithm even better and have more accurate results.

Demo: https://www.youtube.com/watch?v=3UqXOiNmiuM&feature=youtu.be

Future Work

Future Requirement Additions – Search History, iOS availability, Main Menu Customization

Project Files

Project Charter: https://drive.google.com/open?id=1MhzljEl6DNcoTVt2wB6UDiXb9gfBmdU4

System Requirements Specification: https://drive.google.com/open?id=17VbVeWkSdWItvHvHSwVFdDfrw0LSjtcj

Architectural Design Specification: https://drive.google.com/open?id=1NTGWVgVj9IrX49rnLH-mkrLFm6HL061v

Detailed Design Specification: https://drive.google.com/open?id=1gCJlajCeWOO25hE3WLJkvcijT0DBZbVE

Poster https://docs.google.com/presentation/d/1TWTtYCTFWFIeTaG6qg1Pbz6aObsZC7GljHAkiEqJoTY/edit?usp=sharing

References

““API Documentation | TensorFlow.” TensorFlow, www.tensorflow.org/api_docs
Camera | Samsung Galaxy S7 and S7 Edge.” The Official Samsung Galaxy Site, www.samsung.com/global/galaxy/galaxy-s7/camera/.
“Cloud Storage – Online Data Storage | Cloud Storage | Google Cloud.” Google, Google, cloud.google.com/storage/.
“Getting Started | API Client Library for Python | Google Developers.” Google, Google, developers.google.com/api-client-library/python/start/get_started.
“Goodthreads Men’s Slim-Fit Short-Sleeve Solid Oxford Shirt with Pocket.” Amazon, Amazon, www.amazon.com/Goodthreads-Slim-Fit-Short-Sleeve-Oxford-Pocket/dp/B077358ZZ4.