Bassets

Graph
Sample Question
The value (lightness) of colors have the a significant impact on perception even if the contrast value is the same
Figure A (Brand Use Case)
Figure B
Figure C (Grayscale B)
Figure D (Adjusted Saturation)
Unknown Source (Gray)
Unknown Sourse
Same Grayscale different color
Same Grayscale different color (Gray)

🐶Inspiration

The namesake of our project was our four-legged family friends' canines because no one understands the power of grayscale like a species that evolved to live without our luxury of RGB-colored sight*. While interviewing people with visual color disabilities, all of our applicants owned a dog, and one held a Basset that interrupted our meeting. We thought it was a fun reference to remember this fun project.

📈What it does

We created a model for understanding the response time to a corresponding grayscale ratio. On the x-axis, we graphed the contrast, while the y-axis estimated the time it took the user to answer correctly. As we collected data, we saw an exponential decay form in quadrant one. The horizontal asymptote approached a number within the margin of error for average human reaction time: we will refer to this line going further as ‘Ht.’ The vertical asymptote approached within the margin of error for the average most negligible contrast humans can perceive we will refer to this line going further as ‘Hc.’ We then looked at famous International, National, Regional, and Local brand logos contrast to predict their recognizability. The findings were surprising; International/National were all clustered comparatively high compared to local brands, while regional were inconclusive. We believe the data science and marketing team in larger companies are influencing this trend as their market research team came to the same conclusion as us; high contrast means more accessible, faster recall. (see figure A). Why does this small change in reaction time make a difference? We want to introduce you to a concept we decided to call advertising in motion, similar to subliminal advertising. The birth of subliminal advertising dates to 1957, “when a market researcher named James Vicary inserted the words "Eat Popcorn" and "Drink Coca-Cola" into a movie. The words appeared for a single frame, allegedly long enough for the subconscious to pick up but too short for the viewer to be aware. The subliminal ads supposedly created an 18.1% increase in Coke sales and a 57.8% increase in popcorn sales.” ("subliminal advertising") Advertising in motion is any time you see a symbol despite your relative speed or movement pattern and is a subcategory under the subliminal advertisement umbrella. For example, what reaction time is needed to read a billboard when traveling 60 mph on a highway and what is the available readable window? A clever marketer could propose to increase contrast and decrease the reaction time needed to read the sign. If the trend implies a higher difference the less time is required to recognize, then why isn’t everything the max contrast of #FFFFFF (white) and #000000 (black)? The answer is simple advertisement is an art, and breaking the rules can garner attention in a world of conformity. The goal is to stand out after all and, if lucky, become as familiar as household brands like other international/National brands. Our second use case of the model is helping people with color blindness correct their RGB color settings to enjoy media. Our first attempt at color correction was inspired by a post by @colorbindorg on Twitter where they made a point about Fifa's lack of color-blind access ability, given 1 in 12 men are color-blind: see figure B. As the picture suggests, the kits, also known as uniforms, can commonly convert to the same grayscale, which is confusing. To better understand this problem, we recruited four people with Achromatopsia to adjust new models for the individual and help create color filters. Only one of our color-blind applicants, who we will call Applicant A, permitted us to share his data under a masked name, but all four were satisfied with the results. Our first concern with their custom model was some people with Achromatopsia could commonly have reaction time and a higher grayscale minimum perception threshold. Luckly Applicant, A could retrieve this information from documentation from his diagnosis. His reaction time was close to average; therefore, no adjustments to Ht were needed. However, he did have an impairment in his ability to distinguish shades of grays; consequently, we needed to move Hc up along with his graph. We then found his minimum distinguishable contrast ratio for a glance. We then changed the red saturation making it darker than the green jersey by that radio, and asked him to rate his experience watching each, and he loved our adjustments. See figure C & D.

☕️How we built it

Stet Up: Outlined a hypnosis and p value.
Corresponded with visually impaired applicants
Designed a survey with unique pictures (made with photoshop and an automation)
Collected Data
Data entry
Cleaned Data
modeled

💁🏼Challenges we ran into

Our biggest challenge was tackling a problem with significant scope and little and low-quality data existing, especially for the application of helping a rare demographic segment of colorblind people, Achromatopsia. As you can imagine finding 4 of these people was very difficult in a 36 hr time frame. These ambitions led us to build a highly specialized test for our use cases and acquire our dataset while accounting for diversity. In addition, the result has two significant limitations shape variety and lack of simulated motion. An early draft design had us mimicking the movement of shoes while walking and running to predict a drop-in response time.

👨🏻‍💻Accomplishments that we're proud of

We are proud to produce a data model with robust use cases; we haven’t begun to scratch the surface of what is possible.

📚What we learned

We are very excited to learn about inclusivity and accessibility from our new vision-impaired friends and appreciate their insights. For example, web accessibility guidelines ask that static text be a 4.5:1 grayscale ratio, although this is rarely upheld. In addition, orange-yellow color blindness is the rarest and receives little help and funding, making it an attractive target for future applications.

🤖What's next for Bassets

We want to share our findings with local companies failing to take advantage of contrast in advertising which large companies leveraged data science to out-compete. In addition, we would like to create a free, automated, and predictive app for finding the perfect filter toggle extension.

“Does subliminal advertising actually work?” BBC, 20 jan. 2015, https://www.bbc.com/news/magazine-30878843