PhotoSleuth

a novel way to teach Watson by collecting crowdsourced tagged images to turn into informative & marketable datasets

MY ROLE + TEAM

PhotoSleuth was a collaborative effort between IBMers from both IBM Commerce and IBM Watson in Dublin, Ohio. Designers, developers, managers, and information developers made up our team of twelve. I was responsible for leading the visual design and overall aesthetic for our app. I worked closely with the user experience team and developers from both business units. We combined our knowledge of Watson, machine learning, gaming, and software design, and we had a lot of fun along the way!

 
 
 
 
 

App Overview

PhotoSleuth is an exciting, interactive mobile application that collects tagged images from users across the globe to create hand annotated datasets. PhotoSleuth offers a social scavenger hunt experience to both "Clients" - image requestors and "Sleuths" - image seekers.

Participants compete to find high quality photographic "evidence" and earn points based on the quality and quantity of images found.

In this fun social setting, hand annotated data is collectively created around the clock through tagged images and verified by a global community of PhotoSleuth users, resulting in an abundance of images and related metadata to educate Watson.

Hand annotated data is the gold standard in machine learning and is highly marketable. Collecting it, however, is a highly manual process and requires a lot of effort to produce. We designed a better solution by streamlining this normally labor intensive process into a few taps on a mobile device, all while playing the role of a PhotoSleuth in a scavenger hunt.

 
 

design Process

PERSONAS

We created personas not only to clearly identify what we were trying to solve, but for whom we were trying to design our app. Two personas were established as a part of our foundation: one for the sleuth and one for the client. My team and I collaborated on the creation of these personas using lots of sticky notes, and the following profiles are a result of our ideation sessions.

 

Trent
the college student

Male
20 years of age
Single
Landscape architecture major
San Francisco, California

Interests
Photography + traveling to capture the perfect moment
Drinking a good cup of green tea
Gaming with friends
Walking his pug named Ruby
Social media, keeping up to date with trends
Playing laser tag
Watching soccer on Sunday mornings

Pain points
Long hours spent in the studio doing project work
Wishes he had more free time to explore the world

Primary need
Unique, fun gaming experience to help him escape
a little bit from his busy, everyday routine and game
creatively and competitively

 

A Day in the Life of Trent

Julia
THE young professional

Female
27 years of age
Single
Psychology degree
Brooklyn, New York

Work life + Interests
Two years experience at a college health center
Always wants to know why
Loves finding connections in the world
Online shopping
Plays on the NYC Young Professional basketball team
with two of her coworkers-turned-BFFs

Pain points
Balancing work + social life
Love / hate relationship with NYC, she misses the
southern hospitality of her hometown

Primary need
Interested to see how people perceive the world
around them, wants a way to play a game that is
intertwined with her interests and career in psychology

 

A Day in the Life of Julia

 

Hills

Next, we crafted our hills. Hills turn users’ needs into project goals, and they helped our team align around a common understanding of the intended outcomes to achieve. Below are our two hills that we focused on while creating wireframes and workflows for our personas.

 

TRENT COMBINES GAMING AND REAL WORLD EXPLORATION BY COMPETING AGAINST OTHER SLEUTHS IN ADVENTUROUS, PHOTO-SEEKING MISSIONS.

JULIA IS ABLE TO LEARN ABOUT OTHERS' THOUGHT PROCESSES AROUND THE WORLD BY INTERPRETING PHOTOS THAT RESULT FROM THE REQUESTS SHE CREATES WITHOUT A HUGE TIME COMMITMENT.

 

Outcomes

PhotoSleuth was an extremely fun project to design. I learned a lot about game design. I learned how to apply my knowledge of machine learning in a different setting than the setting in which I am normally working with Watson. Although our project is still in the works as far as development, I am excited to see where this project will go!