PhotoTag was idealized as a on or off the cloud photo management application with object recognition AI to help in identifying and finding photos amongst the user's entire library. 

On top of being the lead UI/UX and Design coordinator, I also aided in the programming of the android version of the app. I mainly dealt with the single photo view screen but also helped in the calendar, query, and automated tagging functionality of the app. In accordance with good version control my team and I's contributions to the app can be seen at  the projects GitHub repository .

The app used firebase servers and photo recognition tools to handle account creation and tag suggestions. The asynchronous nature of the server as well as the photo recognition results made for a slight learning curve in their utilization but in the end proved to be worth the effort. On top of the suggestions firebase was able to provide, a member of the iOS team was about to utilize ML databases and tools to create additional, more precise tag suggestions.

As the course project in Smart Sensing and Devices  at Temple University my team and I chose to tackle the challenge of localization to aid the quality of responses from smart things. Only able to focus on creating a demo of the app due to time restraints we used a static UI that depicted a top down view of a team member's home and displayed the recorded information to the user. 

Using the record button the user was able to record the sensing information for the room they are in, this would be repeated for each room for the most accurate readings. Once all the rooms had been recorded the Track button could then be used to start running a program that would predict which room the user was in based on the recorded data. 

Compass and accelerometer information was used in recording Wi-Fi data while the user moves around a room. Knowing that the user could only move from a room to an adjacent room reduced the possible number of room predictions down to a fraction of the total. Using the Wi-Fi data recorded for the different rooms the user's location down to a specific room was predicted.

Though my team and I were able to get the demo to work properly, there were many improvements and additions that we hoped to include in the app including the use of ML for room prediction and a home building feature to create a variety of room configurations. Our Submitted code can be found here.