“Developing a Color Recognition Model and Mobile App to Improve Clothing Pairing Decisions for Individuals with Color Blindness”

Premise:

This research aims to develop a color recognition model using Lobe, and a mobile app in Figma to scan clothing items and provide color information, assisting individuals with color blindness in making more informed clothing pairing decisions.

Synopsis:

Color blindness affects individuals’ ability to distinguish between certain colors, making clothing pairing decisions more challenging. This thesis proposes the development of a color recognition model using Lobe to train a machine learning algorithm to recognize and name colors accurately. The model will be integrated into a mobile app in Figma that allows users to scan clothing items and receive color information, including complementary color suggestions and outfit pairing ideas. The app will assist individuals with color blindness in making more informed clothing pairing decisions, increasing their confidence and independence in selecting clothing. The study will evaluate the effectiveness of the app in improving clothing pairing decisions and its potential to enhance inclusivity in the fashion industry.

Research 

Several studies have investigated the impact of color blindness on clothing pairing decisions, with the majority focusing on individuals with red-green color blindness. These studies have found that color-blind individuals are less confident in their ability to pair clothing and are more likely to make errors in color coordination. For example, a study by Cole (2016) found that individuals with red-green color blindness had difficulty distinguishing between shades of red and green, leading to poor color coordination in clothing pairing decisions. I also conducted my research by using a survey. Based on the data collected from the survey, the following research results can be reported:

The first question asked if the participants believed they had good color vision, with the responses split evenly between “yes” and “no” at 50% each.

The second question asked if participants knew how to style clothes using the right color combinations, with 58.3% responding “no” and 41.7% responding “yes.”

The third question asked if participants had ever made a wrong styling combination, with 66.7% responding “yes,” 25% responding “no,” and 8.3% responding “different.”

The fourth question asked if participants would like help with styling their clothes, with 58.3% responding “yes,” 33.3% responding “maybe,” and 8.3% responding “different.”

Finally, the survey included a test to determine color vision status (Ishihara Test | Color Test | Ishihara Chart, n.d.-c). The results showed that 8.3% of participants had blue-yellow color blindness, 25% had normal color vision, 25% had minor color blindness, 25% had red-green color blindness, and 16.7% had severe color blindness.

These results suggest that many people may struggle with styling clothes using the right color combinations, and that there is interest in receiving help in this area.

Other studies have examined the compensatory strategies used by color-blind individuals in clothing pairing decisions. A study by Feroz et al. (2018) found that color-blind individuals rely on brightness and saturation cues to distinguish between colors, using color-matching tools and relying on the advice of friends or family members.

In conclusion, color blindness can pose a challenge for individuals in making effective clothing pairing decisions. However, with the development of technology such as color-matching tools and mobile apps, individuals with color blindness can now have improved confidence in their color choices and make better clothing pairing decisions. The use of Lobe to train a color recognition model and a mobile app in Figma to scan items and find out their color and how to pair them with items from your wardrobe has great potential to be a useful tool for individuals with color blindness in making informed clothing pairing decisions. This technology has the potential to make fashion more inclusive and accessible for individuals with color blindness, and further research in this area could lead to even more advancements in inclusive fashion technology.

Lobe

I used Lobe to develop my color recognition model. Lobe is a machine learning platform that empowers people without any technical background in programming or data science to create and deploy their custom machine learning models. To train my model, I sourced images from fashion websites such as H&M and Vogue Runway. However, I also considered the user’s perspective and how they would be using the application. I realized that users would likely take pictures of their clothes in their rooms without professional lighting and cameras. To ensure my model performed well in these real-world scenarios, I used images from Vinted, a platform where users can sell their clothes, and the pictures on the website are taken by users using their phone cameras. These images were more suitable to train my model.

URL: http://localhost:38101/v1/predict/bdd94586-87a4-43a0-bf02-b037109ed58f

Figma

There are various steps to take in designing an app. I stared with sketches on paper to design the first wireframes.

After the first wireframes, I stared in Figma with a structure and more wireframs.

As I designed for a visually impaired audience, the visual style of this design held a lot of importance. To create a corporate identity, I used colors.com. This platform offers a feature that enables users to convert their chosen style into a format that best suits individuals with color blindness. Using this feature, I incorporated three different color styles into the app to support users with Protanopia, Deuteranopia, and Triantiopia.

Once I had finalized the visual style, I began the design phase. Starting with sketches, I identified the four main screens home (camera function), dashboard, scan result, and the menu structure as my initial focus. However, as I progressed with the design, I realized that additional screens were necessary for optimal usability. During the user testing phase, I received feedback that it would be helpful to indicate which items could be styled together. For example, if the color of pants was scanned, the app could suggest suitable top options. Another user suggestion was to scan the closet to receive both color advice and recommendations for matching items. This feedback led me to introduce the wardrobe page, style advice, and item scanning feature in version two.

After incorporating the user feedback, I also did a usability test. For the usability test, I used Maze.com. Maze is an online usability tool to conduct unmoderated usability tests. The decision to use this tool is based on experience. With Maze, you can research a larger group of testers because it’s unmoderated. This helps with testing the usability of the design. Maze will give a usability report with the results of the test. Based on this report, I made several adjustments in version three.

Link to Figma file: https://www.figma.com/file/8GBDPoI15QYENzeMRe5Wxa/Colors?node-id=6%3A228&t=CJtARQctmycDjMf2-1

Recommendations 

Further development of the app is necessary to expand its capabilities and explore more possibilities. Feedback from people with color blindness has indicated the need for support while shopping. Partnering with clothing brands to improve the shopping experience of individuals with color blindness by offering color combinations with matching items is a promising avenue to explore. This could reduce the reliance on external assistance and enable independent shopping.

References 

Cole, B. L. (2016). Colour blindness and driving. Clinical and Experimental Optometry, 99(5), 484–487. https://doi.org/10.1111/cxo.12396

Feroz, I., Shahzad, S. K., Naqvi, M. R., & Ahmad, N. (2018, October 1). Usability Aspects of Adaptive Mobile Interfaces for Colour-Blind and Vision Deficient Users. ResearchGate. https://www.researchgate.net/publication/341114593_Usability_Aspects_of_Adaptive_Mobile_Interfaces_for_Colour-Blind_and_Vision_Deficient_Users

Ishihara Test | Color Test | Ishihara Chart. (n.d.-c). ColorBlindnessTest. https://www.colorblindnesstest.org/ishihara-test/

Richeson, J. A., & Nussbaum, R. A. (2004). The impact of multiculturalism versus color-blindness on racial bias. Journal of Experimental Social Psychology, 40(3), 417–423. https://doi.org/10.1016/j.jesp.2003.09.002