-
Context Matters for Image Descriptions for Accessibility: Challenges for Referenceless Evaluation Metrics
Authors:
Elisa Kreiss,
Cynthia Bennett,
Shayan Hooshmand,
Eric Zelikman,
Meredith Ringel Morris,
Christopher Potts
Abstract:
Few images on the Web receive alt-text descriptions that would make them accessible to blind and low vision (BLV) users. Image-based NLG systems have progressed to the point where they can begin to address this persistent societal problem, but these systems will not be fully successful unless we evaluate them on metrics that guide their development correctly. Here, we argue against current referen…
▽ More
Few images on the Web receive alt-text descriptions that would make them accessible to blind and low vision (BLV) users. Image-based NLG systems have progressed to the point where they can begin to address this persistent societal problem, but these systems will not be fully successful unless we evaluate them on metrics that guide their development correctly. Here, we argue against current referenceless metrics -- those that don't rely on human-generated ground-truth descriptions -- on the grounds that they do not align with the needs of BLV users. The fundamental shortcoming of these metrics is that they do not take context into account, whereas contextual information is highly valued by BLV users. To substantiate these claims, we present a study with BLV participants who rated descriptions along a variety of dimensions. An in-depth analysis reveals that the lack of context-awareness makes current referenceless metrics inadequate for advancing image accessibility. As a proof-of-concept, we provide a contextual version of the referenceless metric CLIPScore which begins to address the disconnect to the BLV data. An accessible HTML version of this paper is available at https://elisakreiss.github.io/contextual-description-evaluation/paper/reflessmetrics.html
△ Less
Submitted 27 October, 2022; v1 submitted 21 May, 2022;
originally announced May 2022.
-
Twin-Boundary Structural Phase Transitions in Elemental Titanium
Authors:
Mohammad S. Hooshmand,
Ruopeng Zhang,
Yan Chong,
Enze Chen,
Timofey Frolov,
David L. Olmsted,
Andrew M. Minor,
Mark Asta
Abstract:
Twinning in crystalline materials plays an important role in many transformation and deformation processes, where underlying mechanisms can be strongly influenced by the structural, energetic and kinetic properties of associated twin boundaries (TBs). While these properties are well characterized in common cases, the possibility that TBs can display multiple complexions with distinct properties, a…
▽ More
Twinning in crystalline materials plays an important role in many transformation and deformation processes, where underlying mechanisms can be strongly influenced by the structural, energetic and kinetic properties of associated twin boundaries (TBs). While these properties are well characterized in common cases, the possibility that TBs can display multiple complexions with distinct properties, and phase transitions between them, has not been widely explored, even though such phenomena are established in a few more general grain boundaries. We report experimental findings that {11-24} TBs in titanium display a thick interfacial region with crystalline structure distinct from the bulk. First-principles calculations establish that this complexion is linked to a metastable polymorph of titanium, and exhibits behavior consistent with a solid-state wetting transition with compressive strain, and a first-order structural transition under tension. The findings document rich TB complexion behavior in an elemental metal, with important implications for mechanical behavior and phase-transformation pathways.
△ Less
Submitted 20 July, 2021; v1 submitted 10 March, 2021;
originally announced March 2021.
-
On Computing Average Common Substring Over Run Length Encoded Sequences
Authors:
Sahar Hooshmand,
Neda Tavakoli,
Paniz Abedin,
Sharma V. Thankachan
Abstract:
The Average Common Substring (ACS) is a popular alignment-free distance measure for phylogeny reconstruction. The ACS can be computed in O(n) space and time, where n=x+y is the input size. The compressed string matching is the study of string matching problems with the following twist: the input data is in a compressed format and the underling task must be performed with little or no decompression…
▽ More
The Average Common Substring (ACS) is a popular alignment-free distance measure for phylogeny reconstruction. The ACS can be computed in O(n) space and time, where n=x+y is the input size. The compressed string matching is the study of string matching problems with the following twist: the input data is in a compressed format and the underling task must be performed with little or no decompression. In this paper, we revisit the ACS problem under this paradigm where the input sequences are given in their run-length encoded format. We present an algorithm to compute ACS(X,Y) in O(Nlog N) time using O(N) space, where N is the total length of sequences after run-length encoding.
△ Less
Submitted 16 May, 2018;
originally announced May 2018.
-
Video-Based Facial Expression Recognition Using Local Directional Binary Pattern
Authors:
Sahar Hooshmand,
Ali Jamali Avilaq,
Amir Hossein Rezaie
Abstract:
Automatic facial expression analysis is a challenging issue and influenced so many areas such as human computer interaction. Due to the uncertainties of the light intensity and light direction, the face gray shades are uneven and the expression recognition rate under simple Local Binary Pattern is not ideal and promising. In this paper we propose two state-of-the-art descriptors for person-indepen…
▽ More
Automatic facial expression analysis is a challenging issue and influenced so many areas such as human computer interaction. Due to the uncertainties of the light intensity and light direction, the face gray shades are uneven and the expression recognition rate under simple Local Binary Pattern is not ideal and promising. In this paper we propose two state-of-the-art descriptors for person-independent facial expression recognition. First the face regions of the whole images in a video sequence are modeled with Volume Local Directional Binary pattern (VLDBP), which is an extended version of the LDBP operator, incorporating movement and appearance together. To make the survey computationally simple and easy to expand, only the co-occurrences of the Local Directional Binary Pattern on three orthogonal planes (LDBP-TOP) are debated. After extracting the feature vectors the K-Nearest Neighbor classifier was used to recognize the expressions. The proposed methods are applied to the videos of the Extended Cohn-Kanade database (CK+) and the experimental outcomes demonstrate that the offered techniques achieve more accuracy in comparison with the classic and traditional algorithms.
△ Less
Submitted 5 March, 2015;
originally announced March 2015.
-
A Brief History of Web Crawlers
Authors:
Seyed M. Mirtaheri,
Mustafa Emre Dinçktürk,
Salman Hooshmand,
Gregor V. Bochmann,
Guy-Vincent Jourdan,
Iosif Viorel Onut
Abstract:
Web crawlers visit internet applications, collect data, and learn about new web pages from visited pages. Web crawlers have a long and interesting history. Early web crawlers collected statistics about the web. In addition to collecting statistics about the web and indexing the applications for search engines, modern crawlers can be used to perform accessibility and vulnerability checks on the app…
▽ More
Web crawlers visit internet applications, collect data, and learn about new web pages from visited pages. Web crawlers have a long and interesting history. Early web crawlers collected statistics about the web. In addition to collecting statistics about the web and indexing the applications for search engines, modern crawlers can be used to perform accessibility and vulnerability checks on the application. Quick expansion of the web, and the complexity added to web applications have made the process of crawling a very challenging one. Throughout the history of web crawling many researchers and industrial groups addressed different issues and challenges that web crawlers face. Different solutions have been proposed to reduce the time and cost of crawling. Performing an exhaustive crawl is a challenging question. Additionally capturing the model of a modern web application and extracting data from it automatically is another open question. What follows is a brief history of different technique and algorithms used from the early days of crawling up to the recent days. We introduce criteria to evaluate the relative performance of web crawlers. Based on these criteria we plot the evolution of web crawlers and compare their performance
△ Less
Submitted 4 May, 2014;
originally announced May 2014.
-
A tabu search algorithm with efficient diversification strategy for high school timetabling problem
Authors:
Salman Hooshmand,
Mehdi Behshameh,
Omid Hamidi
Abstract:
The school timetabling problem can be described as scheduling a set of lessons (combination of classes, teachers, subjects and rooms) in a weekly timetable. This paper presents a novel way to generate timetables for high schools. The algorithm has three phases. Pre-scheduling, initial phase and optimization through tabu search. In the first phase, a graph based algorithm used to create groups of l…
▽ More
The school timetabling problem can be described as scheduling a set of lessons (combination of classes, teachers, subjects and rooms) in a weekly timetable. This paper presents a novel way to generate timetables for high schools. The algorithm has three phases. Pre-scheduling, initial phase and optimization through tabu search. In the first phase, a graph based algorithm used to create groups of lessons to be scheduled simultaneously; then an initial solution is built by a sequential greedy heuristic. Finally, the solution is optimized using tabu search algorithm based on frequency based diversification. The algorithm has been tested on a set of real problems gathered from Iranian high schools. Experiments show that the proposed algorithm can effectively build acceptable timetables.
△ Less
Submitted 12 September, 2013;
originally announced September 2013.
