ACM Transactions on Accessible Computing, Nov 14, 2015
Mobile sign language video conversations can become unintelligible if high video transmission rat... more Mobile sign language video conversations can become unintelligible if high video transmission rates cause network congestion and delayed video. In an effort to understand the perceived lower limits of intelligible sign language video intended for mobile communication, we evaluated sign language video transmitted at four low frame rates (1, 5, 10, and 15 frames per second [fps]) and four low fixed bit rates (15, 30, 60, and 120 kilobits per second [kbps]) at a constant spatial resolution of 320 × 240 pixels. We discovered an "intelligibility ceiling effect," in which increasing the frame rate above 10fps did not improve perceived intelligibility, and increasing the bit rate above 60kbps produced diminishing returns. Given the study parameters, our findings suggest that relaxing the recommended frame rate and bit rate to 10fps at 60kbps will provide intelligible video conversations while reducing total bandwidth consumption to 25% of the ITU-T standard (at least 25fps and 100kbps). As part of this work, we developed the Human Signal Intelligibility Model, a new conceptual model useful for informing evaluations of video intelligibility and our methodology for creating linguistically accessible web surveys for deaf people. We also conducted a battery-savings experiment quantifying battery drain when sign language video is transmitted at the lower frame rates and bit rates. Results confirmed that increasing the transmission rates monotonically decreased the battery life.
The quality of lossy compressed images is often characterized by signal-to-noise ratios, informal... more The quality of lossy compressed images is often characterized by signal-to-noise ratios, informal tests of subjective quality, or receiver operating characteristic (ROC) curves that include subjective appraisals of the value of an image for a particular application. We believe that for medical applications, lossy compressed images should be judged by a more natural and fundamental aspect of relative image quality: their use in making accurate diagnoses. We apply a lossy compression algorithm to medical images, and quantify the quality of the images by the diagnostic performance of radiologists, as well as by traditional signal-to-noise ratios and subjective ratings. Our study is unlike previous studies of the effects of lossy compression in that we consider non-binary detection tasks, simulate actual diagnostic practice instead of using paired tests or confidence rankings, use statistical methods that are more appropriate for non-binary clinical data than are the popular ROC curves, and use low-complexity predictive tree-structured vector quantization for compression rather than DCT-based transform codes combined with entropy coding. Our diagnostic tasks are the identification of nodules (tumors) in the lungs and lymphadenopathy in the mediastinum from computerized tomography (CT) chest scans. Radiologists read both uncompressed and lossy compressed versions of images. For the image modality, compression algorithm, and diagnostic tasks we consider, the original 12 bit per pixel (bpp) CT image can be compressed to between 1 bpp and 2 bpp with no significant changes in diagnostic accuracy. The techniques presented in this paper for evaluating image quality do not depend on the specific compression algorithm and are useful new methods for evaluating the benefits of any lossy image processing technique. IGITAL image processing in recent years has shown D tremendous potential for application to digital medical images. At the foreground are the possibilities for easy image Manuscript
This paper presents an H.264 standard-compliant video encoder optimized for region-of-interest (R... more This paper presents an H.264 standard-compliant video encoder optimized for region-of-interest (ROI) based coding tuned to American Sign Language (ASL) videos. Encoding modes are developed which allow the encoder to allocate both rate and computational resources differently across the ROIs. An objective measure of intelligibility is included in an encoder parameter optimization by modifying a fast offline distortion-complexity optimization algorithm, resulting in parameter selections that demonstrate excellent rateintelligibility-complexity performance. These parameters can be stored in a look-up table for use by an online algorithm which selects parameters based on available computational resources. The resulting parameter selections improve the encoder speed by up to 21.2% with a small decrease in intelligibility over the x264 default parameter settings.
Equality, Diversity and Inclusion: An International Journal, Apr 15, 2019
Purpose -The advancement of equity, diversity and inclusion in higher education is dependent on i... more Purpose -The advancement of equity, diversity and inclusion in higher education is dependent on institutional culture changes in academia. Faculty equity, diversity and inclusion efforts must engage departmental leadership. The purpose of this paper is to describe the growth and expansion of the ADVANCE leadership program at the University of Washington (UW) for department chairs that was designed to provide department chairs the skills, community and information needed to be agents of change within the academy. Design/methodology/approach -The paper chronicles the program's growth from a campus-based workshop program to national workshops (LEAD) to a web-based toolkit (LiY!) to support institutions in running their own UW ADVANCE-inspired leadership workshops. Findings -The paper demonstrates the success of each growth stage and the expansion of program impact. Practical implications -The paper offers recommendations for growing a model from a local to national scale and adapting the described leadership development model at other institutions. Originality/value -The paper shares a successful model for equipping department chairs to be advocates of gender equity, diversity and inclusion in STEM and to be change agents in higher education.
We describe our system called MobileASL for real-time video communication on the current U.S. mob... more We describe our system called MobileASL for real-time video communication on the current U.S. mobile phone network. The goal of MobileASL is to enable Deaf people to communicate with Sign Language over mobile phones by compressing and transmitting sign language video in real-time on an off-the-shelf mobile phone, which has a weak processor, uses limited bandwidth, and has little battery capacity. We develop several H.264-compliant algorithms to save system resources while maintaining ASL intelligibility by focusing on the important segments of the video. We employ a dynamic skin-based region-of-interest (ROI) that encodes the skin at higher quality at the expense of the rest of the video. We also automatically recognize periods of signing versus not signing and raise and lower the frame rate accordingly, a technique we call variable frame rate (VFR). We show that our variable frame rate technique results in a 47% gain in battery life on the phone, corresponding to an extra 68 minutes of talk time. We also evaluate our system in a user study. Participants fluent in ASL engage in unconstrained conversations over mobile phones in a laboratory setting. We find that the ROI increases intelligibility and decreases guessing. VFR increases the need for signs to be repeated and the number of conversational breakdowns, but does not affect the users' perception of adopting the technology. These results show that our sign language sensitive algorithms can save considerable resources without sacrificing intelligibility.
to reduce the computational complexiry by performing them simultaneously with the compression. Af... more to reduce the computational complexiry by performing them simultaneously with the compression. After briefly reviewing the fundamental ideas of vector quantization, we present a survey of vector quantization algorithms that perform image processing.
The goal of the MobileASL (American Sign Language) research project is to enable sign language co... more The goal of the MobileASL (American Sign Language) research project is to enable sign language communication over the U.S. cellular network, which is low bandwidth and lossy. Data loss can greatly impact the quality of compressed video because of temporal and spatial error propagation. We investigate techniques to minimize the effect of data loss for improving compressed video sign language conversations. As both computational power and bandwidth are limited on cellular devices, we must carefully allocate these resources. Specifically we focus on utilizing feedback to recover from data loss.
The primary challenge to enabling real-time twoway video conferencing on a cell phone is overcomi... more The primary challenge to enabling real-time twoway video conferencing on a cell phone is overcoming the limited bandwidth, computation and power. The goal of the MobileASL project is to enable access for people who use American Sign Language (ASL) to an off-the-shelf mobile phone through the implementation of real-time mobile video communication. The enhancement of processor, bandwidth, and power efficiency is investigated through SIMD optimization; region-of-interest encoding based on skin detection; video resolution selection (used to determine the best trade off between frame rate and spatial resolution); and variable frame rates based on activity recognition. Our prototype system is able to compress, transmit, and decode 12-15 frames per second in real-time and produce intelligible ASL at 30 kbps. Furthermore, we can achieve up to 23 extra minutes of talk time, or a 8% gain over the battery life of the phone, through our frame dropping technique.
Acta Crystallographica Section D-biological Crystallography, Feb 22, 2006
Protein crystallography labs are performing an increasing number of experiments to obtain crystal... more Protein crystallography labs are performing an increasing number of experiments to obtain crystals of good diffraction quality. Better automation has enabled researchers to prepare and run more experiments in a shorter time. However, the problem of identifying which experiments are successful remains difficult. In fact, most of this work is still being done manually by humans. Automating this task is therefore an important goal. As part of a project to develop a new and automated high-throughput, capillary-based protein crystallography instrument, we have been developing a new image classification subsystem to greatly reduce the number of images that require human viewing. This system must have low rates of false negatives (missed crystals), possibly at the cost of raising the number of false positives. The image classification system employs a support vector machine (SVM) learning algorithm to classify the blocks of each image. A new algorithm to find the area within the image that contains the drop is employed. The SVM uses numerical features, based on texture and the Gabor wavelet decompostion, that are calculated for each block. If a block within an image is classified as containing a crystal, then the entire image is classified as having a crystal. In a study of 375 images, with 87 containing crystals, a false negative rate of less than 4%, with a false positive rate of about 40% is consistently achieved.
is the associate dean for undergraduate programs in the College of Engineering and Architecture a... more is the associate dean for undergraduate programs in the College of Engineering and Architecture at Washington State University (WSU) and a faculty member in WSU's School of Electrical Engineering and Computer Science (EECS). He has taught courses ranging from introductory programming at the freshman level to advanced electromagnetics at the graduate level. He is a Fellow of the IEEE and has been selected as the WSU EECS Researcher of the Year and the School's Teacher of the Year.
In presenting this dissertation in partial fulfillment of the requirements for the doctoral degre... more In presenting this dissertation in partial fulfillment of the requirements for the doctoral degree at the University of Washington, I agree that the Library shall make its copies freely available for inspection. I further agree that extensive copying of this dissertation is allowable only for scholarly purposes, consistent with "fair use" as prescribed in the U.S.
Many classes of images possess a strong degree of spatial stationarity such that particular featu... more Many classes of images possess a strong degree of spatial stationarity such that particular features of the image reliably appear in certain regions of the image. This spatial information can be used to improve compression. Unbalanced or Pruned Tree-structured Vector Quantization (PTSVQ) is a variable-rate coding technique that tends to use more bits to code active regions of the image, and fewer to code homogeneous ones. The P T S V Q is developed based on a training sequence of typical images. We used a regression tree algorithm to segment the images of the training sequence, using the t, y pitel location as a predictor f o r the intensity. This segmentation was used t o partition the training data by region and generate separate codebooks for each region, and to allocate diflering numbers of bits to the regions. Unlike other varieties of classzfied vector quantization, a region-based classification requires no side information as the decoder kiiows where in the image the current encoded block originated. These methods can enhance the perceptual quality of compressed images when compared with ordinary PTSVQ. Results are shown on magnetic resonance data.
The current recommended video transmission standards, Telecommunication Standardization Sector (I... more The current recommended video transmission standards, Telecommunication Standardization Sector (ITU-T) Q.26/16, of 25 frames per second at 100 kilobits per second or higher make mobile sign language video communication less accessible than it could be with a more relaxed standard. The current bandwidth requirements are high enough that network congestion may occur, causing delays or lost information. In addition, limited data plans may cause higher cost to video communication users. To increase the accessibility and affordability of video communication, we explore a relaxed standard for video transmission using lower frame rates and bitrates. We introduce a novel measure, the Human Signal Intelligibility Model, to accomplish this. We propose web and laboratory studies to validate lower bounds on frame rates and bitrates for sign language communication on small mobile devices. Author
She works on the evaluation of several projects aimed at improving diversity, equity, and inclusi... more She works on the evaluation of several projects aimed at improving diversity, equity, and inclusion in STEM fields. She also conducts research on the socialpsychological and institutional forces that contribute to the persistence of race and class inequalities in the United States. Emily earned a PhD and MA in Sociology from the University of Washington, and a BA in Sociology from Smith College.
She investigates the historical and cultural dimensions of underrepresented groups' participation... more She investigates the historical and cultural dimensions of underrepresented groups' participation in science, technology and engineering and the reasons why white males still dominate these fields.
2018 ASEE Annual Conference & Exposition Proceedings
Boulder. She holds a Ph.D. in Engineering and Public Policy from Carnegie Mellon University and a... more Boulder. She holds a Ph.D. in Engineering and Public Policy from Carnegie Mellon University and a J.D. from the University of Colorado School of Law. Her research and teaching focus on atmospheric chemistry and transport modeling and air quality management.
Mobile sign language video conversations can become unintelligible if high video transmission rat... more Mobile sign language video conversations can become unintelligible if high video transmission rates cause network congestion and delayed video. In an effort to understand the perceived lower limits of intelligible sign language video intended for mobile communication, we evaluated sign language video transmitted at four low frame rates (1, 5, 10, and 15 frames per second [fps]) and four low fixed bit rates (15, 30, 60, and 120 kilobits per second [kbps]) at a constant spatial resolution of 320 × 240 pixels. We discovered an “intelligibility ceiling effect,” in which increasing the frame rate above 10fps did not improve perceived intelligibility, and increasing the bit rate above 60kbps produced diminishing returns. Given the study parameters, our findings suggest that relaxing the recommended frame rate and bit rate to 10fps at 60kbps will provide intelligible video conversations while reducing total bandwidth consumption to 25% of the ITU-T standard (at least 25fps and 100kbps). As...
Uploads
Papers by Eve Riskin