Update (March 8, 2014): My research was featured on several other sites, including news websites from China, Spain, Croatia and France. I have been receiving a lot of positive feedback and I’m very excited to hear from the large number of people whose lives this project could touch!
Update (2 March 2014): Funding goal reached! Thank you for the contributions. Contributions beyond the initial goal will be used to collect an even larger number of samples for the research - a larger sample size will directly lead to a more optimal algorithm.
Update (1 March 2014): My research was featured on Mashable! Special thanks to Melissa Goldin for getting in touch :)
Update (15 February 2014): Thanks to very, very generous contributions, this project has almost reached it's funding goal within two days. I cannot begin to describe how helpful this contribution would be in my performance at the Intel ISEF and my motivation for research. Thank you very much for the support! Further contributions would be used to substantiate the survey's sample size - a larger sample size of color-blind users would lead to more refinement and improvements in the color-correction algorithms.
Around 7% of the world's population is color-blind. This Intel International Science and Engineering Fair research project is based on color compensation algorithms and techniques that improve images for color-blind viewers.
I'm Animesh, 17, a grade 12 student from New Delhi, India. Over the last two years, I have been working on a research project that aims to optimize images, videos, websites and other visual content for color-blind viewers using image-processing algorithms. This research explores four image optimization algorithms with several permutations of color-correction techniques, and aims to determine the best color-correction process.
In June 2013, this research won recognition when I was selected as a Regional Finalist at the Google Global Science Fair. After working throughout my senior year to bring this project to the Intel ISEF, I have been selected to represent India at the international-level fair in May 2014 at Los Angeles, California.
1) In order to extend my research, prove the effectiveness of the algorithms, and determine the most efficient algorithm, I need to test it on a sufficiently large sample set. While I have tested the algorithms with several people in my community, finding a mobilized group of color-blind viewers has been difficult.
However, by leveraging the power of crowdsourcing and the Internet, I can obtain responses to a survey by presenting images optimized using four novel algorithms to color-blind viewers. A larger sample size would lead to increased credibility and a stronger basis for this research. Amazon MechTurks and a social-media application are currently top choices for these surveys.
2) I have currently been using unlicensed software. However, for the international level fair, I have been instructed to purchase licensed copies of MATLAB, Mathematica and other digital image processing toolkits that I am using in my research.
3) I am currently using my own web-server that has several bandwidth and speed constraints. As a result, even relatively simple computations take upto 10 seconds of processing time. In order to develop a scalable and universal application for improving images, I need to use Amazon Web Services or an equivalent host that is capable of handling a multitude of image processing requests at any given time.
With sufficient data and the right platform, I believe this project could immensely benefit the color-blind community. Over the next few years, with calibration and improvements, these optimization algorithms could find their way into devices as software-based accessibility features.
- Amazon MechTurks is a crowdsourcing marketplace where real people perform surveys and other HITs (Human Intelligence Tasks) to earn rewards of the order of 1 or 2 cents for each task. This platform can be leveraged to find color-blind people and test them for color-blindness, allowing them to determine the best perceivable images from a variety of samples generated using the enhancement algorithms.
- More funds would lead to more volunteers for the color-blindness tests. During manual surveys, only one in every twelve volunteers was diagnosed with color blindness. Thus, one volunteer would not translate to one sample at most times, and a significant number of people would need to be tested before a sufficient data set is created.
- A portion of the funds would be used to buy licenses and hosting for the project source code. This is an important step before the project can be exhibited at the international level.
I am still a high school student, so I cannot offer a lot of fancy rewards, but I would like to show my appreciation in the following ways:
- For contributions upto $25, the contributors names would be listed on the acknowledgements in my project report at the Intel ISEF.
- For contributions upto $50, the contributors would receive an additional handwritten Thank You card from me in addition to the above.
- For contributions upto $100, the contributors names would receive a special acknowledgement on the project's webpage in addition to the above rewards.
- For contributions upto $500, the contributors names would be mentioned on my main project panel at the Intel ISEF.
The motivation behind this project was to help color-blind people perceive information that they would otherwise be missing. Color-blindness is often left undetected, and even when detected, people rarely visit doctors for this condition.
During the course of my research, I visited several blind schools, ophthalmologists and vision centers, but I could never gather information about people suffering from color-blindness. I then used simulation algorithms to predict the most effective methods for color compensation for color-blind viewers.
Testing can corroborate the findings and developments of my research, and a proven algorithm could then be used as the norm for color-correction. This feature could then be built into any device - phones, tablets, desktops, billboards - every element with a visual or digital aspect would benefit from this research.
With your help, I can substantiate this research and present a viable solution for the millions of color blind viewers across the world.
Other Ways You Can Help
Please help share this campaign so I can gather more support for the research!
Note: All logos used above are copyrights of their respective owners.