Our Findings

Accessible Code required extensive research about visual impairment both before and throughout the development process. This is the background story of how it came to be.

How the project started

In August 2016, PIJIN, an affiliate company of Export Japan, received an inquiry from Kobe Light House. This is an organization that helps visually impaired people (VIP) live their lives more independently.

PIJIN has a QR code-based product called QR Translator for multilingual use, and Kobe Light House wondered whether its text-to-speech function could be used by VIP to obtain information from printed materials.

From that point on, we began studying visual impairment and the needs of VIP.

An illustration of two people using smartphones and desktop computers, with Accessible Code user interface in the background.

Is braille a solution?

Before starting the Accessible Code project, we were under the impression that braille was universal. Unfortunately, this is not the case. Because braille corresponds to different sounds in different languages, every language needs different braille.

The Japanese Braille alphabet and English Braille alphabet are side by side for comparison, showing the same braille alphabet may refer to a different sound or letter in each language.

Braille is indispensable for a large number of people with visual impairments, and we are in no way contesting its importance. However, only about one-eighth of VIP can read braille, and braille cannot easily fit on small or limited spaces, such as product labels. This means there is still a need for more solutions.

Smartphone accessibility

Thanks to the efforts of developers who have come before us, most modern smartphones have many different accessibility features that support VIP in their daily lives. One of the most useful features is the screen reader. This is called “VoiceOver” on the iOS platform and “TalkBack” on the Android platform.

Three smartphone screens show the steps for setting up VoiceOver on iOS, with the menu items Accessibility, VoiceOver, and VoiceOver with a toggle setting outlined in red boxes.
Three smartphone screens show the steps for setting up TalkBack on Android, with the menu items Accessibility, TalkBack, and TalkBack with a toggle setting outlined in red boxes.

Thanks to features like these, more and more VIP, including fully blind people, have started using smartphones and other similar information and communications technology (ICT).

Note: Data sourced from “An analysis of differences among residences in the usage of communication support services and ICT devices by blind and visually impaired people” (Tetsuya Watanabe, Faculty of Engineering, Niigata University, 2017).

A bar chart shows the results of a survey of blind people about their use of information and communications technology in 2013 and 2017. The percentage of those who used regular mobile phones dropped from 85.8% to 60.9%. For those who used smartphones and tablets, the share increased from 22.6% to 52.1% and from 9.5% to 14.4%, respectively. Almost everyone used personal computers in both years, with 96.3% in 2013 and 96.7% in 2017.A bar chart shows the results of a survey of blind people about their use of information and communications technology in 2013 and 2017. The percentage of those who used regular mobile phones dropped from 85.8% to 60.9%. For those who used smartphones and tablets, the share increased from 22.6% to 52.1% and from 9.5% to 14.4%, respectively. Almost everyone used personal computers in both years, with 96.3% in 2013 and 96.7% in 2017.

Why QR codes?

We learned from VIP and their supporting organizations that some VIP already use QR codes in their daily life. We then hypothesized that QR codes could be an alternative to braille if VIP could scan QR codes with their smartphones without any difficulty.

QR code was invented by Denso, a global company headquartered in Japan, and it is a patent-free universal technology already prevalent worldwide. QR code scanning applications are built into many modern smartphones by default.

However, when we started this project, we conducted research in multiple languages but found no data on what percentage of VIP could actually use QR codes. There was also a lack of information on what conditions need to be met to make the codes easy to locate and use.

That is why we decided to obtain this data ourselves, in cooperation with organizations that support VIP.

Consultation, monitoring, and experiments

We needed quantitative scientific data to prove our hypothesis, but the cost was more than we could shoulder alone. We sought support from a Japanese government agency, New Energy and Industrial Technology Development Organization (NEDO), for their Development Promotion Project for Practical Use of Welfare Equipment subsidy program.

Fortunately, we were qualified to receive the subsidy, and this kicked off our nationwide research. With the help of organizations that support VIP, we asked 150 VIP to take part in consultations and experiments that would help us determine the best conditions for developing helpful QR codes.

These are some of the key results of the consultations, monitoring, and experiments.

An illustration of a person with a tablet and a person with a desktop computer and books, with Accessible Code user interface in the background.

Needs of VIP

The first step was to ask VIP what their actual needs are in their everyday routines. There were responses from 100 VIP who had various kinds of visual impairment. They told us what kind of information they want to obtain through QR codes, especially based on what kind of information they feel is difficult to access in their daily lives.

These were the results:

A bar chart shows 50 responses for food product types, 35 for food expiration dates, 22 for beverage types, 19 for place of origin and list of ingredients, 17 for prescription medicine information, 16 for clothing information, 14 for types of detergents, 12 for postal information, 12 for how to cook instant noodles, 9 for bus and train timetables, 8 for restaurant menus, 7 for electrical product manuals, 6 for documents from government offices, and 5 for types of music CDs.A bar chart shows 50 responses for food product types, 35 for food expiration dates, 22 for beverage types, 19 for place of origin and list of ingredients, 17 for prescription medicine information, 16 for clothing information, 14 for types of detergents, 12 for postal information, 12 for how to cook instant noodles, 9 for bus and train timetables, 8 for restaurant menus, 7 for electrical product manuals, 6 for documents from government offices, and 5 for types of music CDs.

There were a lot of concerns about the safety of products to ingest, like food and medicine.

Can VIP scan QR codes?

What was most important in our first experiment was to determine whether or not VIP can easily scan QR codes on plain paper by using smartphones in a room with sufficient lighting.

Throughout the first experiment, it became clear that VIP could scan QR codes without difficulty if they had a chance to get used to doing so. We also learned that some of the experienced QR code users used similar hand positions, as indicated in the picture below, when they successfully scanned QR codes.

This was considered the most effective posture for VIP to scan QR codes.

A person’s hand is holding a smartphone approximately 20 to 30 centimeters above a desk, where there is a flier with a QR code on it.
A bar chart shows the success rate of code scanning before guidance for 100 participants, who had an average of 52 percent success. Blind participants had 36.7 percent success, low vision participants had 75 percent success, smartphone users had 71.4 percent success, and QR code users had 71.8 percent success.A bar chart shows the success rate of code scanning before guidance for 100 participants, who had an average of 52 percent success. Blind participants had 36.7 percent success, low vision participants had 75 percent success, smartphone users had 71.4 percent success, and QR code users had 71.8 percent success.A bar chart shows the success rate of code scanning after guidance for 100 participants, who had an average of 98 percent success. Blind participants had 98.3 percent success, low vision participants had 97.5 percent success, smartphone users had 100 percent success, and QR code users had 100 percent success.A bar chart shows the success rate of code scanning after guidance for 100 participants, who had an average of 98 percent success. Blind participants had 98.3 percent success, low vision participants had 97.5 percent success, smartphone users had 100 percent success, and QR code users had 100 percent success.

This video shows them in action as they scan the codes.

What are the factors that affect how VIP scan QR codes?

In our following experiments, we tested many different factors we assumed might have an effect on how easily VIP could scan QR codes. These are the factors that we found most affected the code scanning success rate.

The first factor is the size of the code. Like mentioned above, many VIP scan the codes from about 20 to 30 cm away, so the size of the code has to be big enough for their camera to read. We have concluded that dimensions of at least 10 mm squared are ideal, though this is somewhat affected by different camera resolutions and different versions of QR codes. 

A bar chart shows the success rate by code size for 50 participants. At 0.6 centimeters, 48 percent had success, at 0.8 centimeters, 82 percent, at 1 centimeter, 96 percent, and at 1.2 centimeters, 100 percent.A bar chart shows the success rate by code size for 50 participants. At 0.6 centimeters, 48 percent had success, at 0.8 centimeters, 82 percent, at 1 centimeter, 96 percent, and at 1.2 centimeters, 100 percent.

The second factor is the tactile markings that indicate the location of the code. VIP, especially blind ones, have no visual clues to identify the location of QR codes. This was not a big issue in the case of QR codes on plain sheets of paper, but in cases of 3-dimensional objects like consumer goods, tactile markings became a more significant factor.

There were not many differences among the types of tactile markings. However, the presence of a tactile marking seemed to be an essential factor for blind people to scan the code after identifying its location.

A bar chart shows the success rate for 50 participants using different tactile marks. With dots, about 95 percent had success, with seal stickers, about 95 percent, with concave marks, roughly 98 percent, with convex marks, roughly 95 percent, and with no tactile marks, roughly 68 percent.A bar chart shows the success rate for 50 participants using different tactile marks. With dots, about 95 percent had success, with seal stickers, about 95 percent, with concave marks, roughly 98 percent, with convex marks, roughly 95 percent, and with no tactile marks, roughly 68 percent.

The video to see how tactile marks work.

You can refer to our summarized report to learn more about the experiments we conducted.

Above all, these experiments were a valuable opportunity for us to learn a lot by spending time with VIP and their supporting organizations. Their input is ultimately what led to the birth of Accessible Code.

Looking for more information?

There’s still lots more to learn about Accessible Code! Want to know more about the cost structure, how the content is translated into multiple languages, and how visually impaired people use the code? Please take a look at the FAQ page.

Visit Frequently Asked Questions