Students respond to a multiple-choice question by holding up one of four printed cards.

For a concise news article on this project, see the MSR cover story by Douglas Gantenbein.

INTRO

We present here a low-cost alternative to electronic response systems (i.e. i>clicker) using a laptop, webcam, and paper cards to use for voting. The system was originally designed to be used by a teacher polling students in a classroom on multiple-choice questions, and the description below is of our initial work in this area. However, the system can be easily adapted to other scenarios such as market research, audience choice awards, conference surveys, and anonymous polls for sensitive topics.

Moving forward, we are looking to expand our work in the education space by doing further trials and get the system working on a mobile phone platform. We are also interested in experimenting with other applications mentioned above and would appreciate any feedback or other ideas where this design might have an impact.

see pictures: polling 1 | camera setup | camera perspective of poll

watch videos: polling

further reading: UIST 2012 submission on qCards

PROJECT SUMMARY

Electronic response systems known as “clickers” have demonstrated educational benefits in well-resourced classrooms, but remain out-of-reach for most schools due to their prohibitive cost.

We propose a new, low-cost technique that utilizes computer vision for real-time polling of a classroom. Our approach allows teachers to ask a multiple-choice question. Students respond by holding up a sheet of paper that contains a printed code, similar to a QR code, encoding their student IDs. By holding the card in different rotation orientations, they can answer in four distinct ways (i.e. A, B, C, or D). Using a laptop and an off-the-shelf webcam, our software automatically recognizes and aggregates the students’ responses and displays them to the teacher. Each student's card pattern and rotation mapping to answer choices is different, so anonymity between students is preserved, though the teacher can still know how each student answered the question.

We built this system and performed initial trials in secondary schools in Bangalore, India. In a typical 25-student classroom, our system offers 99.8% recognition accuracy, captures 97% of responses within 10 seconds, and costs 15 times less than existing electronic solutions.

LARGE AUDIENCES

We have also conducted initial experiments in polling large audiences with this same technique. For larger audiences, a webcam cannot get sufficient resolution to decode the qCards, so a higher-resolution DSLR camera was used.

At the 2012 ACM UIST conference, we conducted a poll with about 270 participants. Our system read 90% of responses, decoding them with 98% accuracy. The images below show a panorama of 5 separate images stitched together of the audience before and after decoding their responses.

    further reading: Supplemental Material on Low-cost Polling of Large Audiences

SYSTEM OVERVIEW

Our system can be instantiated in several different forms. The instantiation used for our experiments is illustrated on the next page. The fundamental elements of our design are as follows:

1. Printed cards for each student. Each card encodes a unique bit pattern indicating the student's ID. Each of the four discrete orientations of the card maps to a multiple-choice answer, (i.e. A, B, C, or D). The answer choices are also written discretely on the back of each card corresponding to each rotation, so that students can clearly see how to hold the card to register a given multiple choice answer. Cards are printed on A4 paper using a normal black-and-white printer.
2. Camera-enabled computing device. In our experiments with smaller audiences, we utilize a laptop with a USB webcam (Logitech Pro 9000, 2.0 megapixel video). We envision that a mobile phone can substitute for the laptop and webcam in the future. For larger audiences, we have also experimented with using a DSLR camera for capturing the images for larger rooms at greater distance.
3. (Optional) LCD projector. If available, an LCD projector can be used to display the results of each poll to the class. We utilized such a projector in our initial prototyping, but not in our final tests, as the government school where we ran our experiment did not have a projector in the classroom. For each question asked by the teacher, the options were written on the board prior to initiating a poll.

APPLICATIONS OUTSIDE OF EDUCATION

While our current focus is on using polling for classroom education, there are many applications in other contexts. Examples include:

• market research
• audience choice awards
• conference surveys
• anonymous polls for sensitive topics

We envision that the technology developed in this paper would apply equally well in these scenarios and are open to further suggestions for applicable scenarios.