Multiple View Perspectives: Improving Inclusiveness and Video Compression in Mainstream Classroom Recordings
Raja S. Kushalnagar, Anna C. Cavender, Jehan-François Pâris · 2010 · Proceedings of the 12th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2010) · doi:10.1145/1878803.1878827
Summary
This paper presents Multiple View Perspectives (MVP), a system that captures and presents multiple focused video views of a classroom for deaf and hard of hearing (DHH) students. DHH students in mainstream classrooms face a fundamental visual attention problem: they must simultaneously watch the instructor, sign language interpreter, real-time captions, presentation slides, and classmates — information sources that are spatially distributed around the room. Since they can only attend to one visual source at a time, they inevitably miss information. MVP addresses this by using multiple inexpensive webcams or mobile phone cameras to capture separate regions of interest (the instructor, interpreter, slides, and captions), then presenting these as focused views on a single laptop screen. The system exploits the different temporal characteristics of each view for video compression — slides change slowly (recompressed to 1 fps, reducing file size by 95%), captions update moderately (10 fps, 66% reduction), and the instructor view changes most rapidly (30 fps, 15% reduction). The combined MVP file was 75% smaller than an equivalent single high-definition video. The paper reports two iterative evaluation rounds with 19 DHH participants (ages 20-45) at the Rochester Institute of Technology, comparing Single View Perspective (SVP), four-view MVP, and a refined Compact MVP (C-MVP) that reduced the layout to three views with the instructor as a picture-in-picture inset.
Key findings
In the first evaluation, participants rated MVP (M=3.8) slightly higher than SVP (M=3.5) but the difference was not statistically significant (chi-squared=19.87, N=19, p=0.019). However, participants found the four-view layout too distracting and noted the instructor view was too large. This feedback led to the C-MVP design with three windows plus a picture-in-picture inset. In the second evaluation, C-MVP was rated significantly higher than both MVP and SVP across all five measures (all p<0.005): overall rating (C-MVP 4.5 vs MVP 3.7 vs SVP 3.5), helpfulness (4.7 vs 4.0 vs 3.6), ease of use (4.4 vs 3.5 vs 3.8), confidence in using (4.6 vs 3.9 vs 3.6), and likelihood of recommending to other DHH students (4.8 vs 4.2 vs 3.6). Participants noted that C-MVP separated all information into fixed locations with no empty space, making it easier to locate relevant content. The approach reduced file size and bandwidth by 75%, enabling streaming to laptops and mobile devices. Importantly, participants who use accommodations reported that the instructor did not need high resolution — a small sub-picture within the interpreter view was sufficient.
Relevance
This research addresses a persistent challenge in inclusive education: how to make classroom presentations visually accessible for DHH students without relying on expensive professional video production. The MVP approach puts control in students' hands through low-cost, portable technology rather than depending on institutional infrastructure or camera operators. For accessibility practitioners, the key design insight is that DHH students prefer multiple focused views over a single high-quality panoramic view — reducing visual dispersion by bringing spatially distributed information sources closer together on screen is more valuable than increasing resolution. The iterative design process, where student feedback transformed the four-view layout into the more effective three-view C-MVP, demonstrates the importance of involving DHH users directly in refining video presentation formats. The compression techniques based on different temporal characteristics of each view remain relevant for modern lecture capture and streaming systems.
Tags: deaf and hard of hearing · lecture capture · video accessibility · classroom accessibility · sign language interpreting · captioning · visual attention · video compression · inclusive education
Standards referenced: H.264