Department of Computer Science
Rochester Institute of Technology
Phone: (585) 475-4536
|[ Home ]||[ Members ]||[ Projects ]||[ Publications ]||[ Software ]||[ Support ]|
The Document and Pattern Recognition Lab (dprl) is located in the Department of Computer Science at the Rochester Institute of Technology in Rochester, New York, USA. Our lab explores ways to improve the algorithms and tools used for automatic recognition and retrieval of information in documents, images, audio and video.
News (2014) (Earlier News)
- (Oct.) A paper co-authored by Chris Riedl (Northeastern, former Harvard post-doc), Marti Hearst (UC Berkeley), Siyu Zhu, Richard Zanibbi and researchers from the Harvard-NASA Tournament Lab (Karim Lakhani et al.) describing an online competition for labeling parts in US patent diagram images has been posted on the arXiv. The data and source code for the top-5 placing systems in the competition are available through the UCI Machine Learning Repository. It has been a challenge to find a home for this paper due to its interdisciplinary nature (despite some very positive reviews), but hopefully an appropriate venue will be found in the near future.
- (Oct.) David Stalnker's paper Math expression retrieval using an inverted index over symbol pairs has been accepted for oral presentation at Document Recognition and Retrieval XXII (2015) in San Francisco.
- (Oct.) Nidhin Pattaniyil extended David's work for his MSc project (paper), adding support for matrices and combining expression retrieval with text search to create an entry for the international NTCIR Math Retrieval Competition. The competition was held for the upcoming NTCIR-11 conference in Japan. David and Nidhin's system performed best for retrieving Wikipedia expressions, and had the best overall top-5 results and second-highest 'highly relevant' top-5 results for combined text and math queries for 100,000 documents from the arXiv. A paper describing the updated system and the competition results will appear in the NTCIR-11 proceedings this December.
- (Sept.) Prof. Zanibbi gave a talk in the Human-Computer Interaction Seminar at the University of Waterloo on Sept. 26th, entitled "Creating User-Friendly Math Search Engines." The talk covered search interface designs and math retrieval algorithms developed in the DPRL. Our thanks to Dr. Edward Lank for the invitation to give this talk.
- (Sept.) The successful ICFHR 2018 Rochester bid has been announced on the GCCIS Web Pages.
- (Sept.) Prof. Zanibbi submitted a successful bid to host the 2018 International Conference on Frontiers in Handwriting Recognition in Rochester, at RIT. The bid is available here, and the presentation slides (presented by Umapada Pal at ICFHR 2014 in Crete) are available here. If you work in handwriting recognition, we hope to see you in 2018!
- (Aug.) Kevin Talmadge has succesfully completed his Master's Project on applying the Recognition Strategy Library to analyzing a math recognition system.
- (June) As part of an Image and document processing, recognition and interaction seminar held at IUT/Univ. Nantes (France, on June 30), Prof. Zanibbi gave a talk on human math retrieval studies and math search engines (work done with Keita Wangari, Matthias Reichenbach, David Stalnaker and Nidhin Pattaniyil.) Other presenters were researchers from Australia, Brazil, and China. (program; slides)
- (June) Amit Pillay has successfully defended his MSc thesis, Intelligent Combination of Structural Analysis Algorithms: Application to Mathematical Expression Recognition, which presents a method for combining DRACULAE math notation parsers using a Graph Transformer Network (GTN). This technique was successful in increasing spatial relationship detection accuracy.
- (May) Congratulations to Kenny Davila, who has succesfully completed his Research Potential Assessment, which is required for PhD students at the end of their first year.
- (May) The lab welcomes Kevin Talmadge, who will be working on the new Python-based Recognition Strategy Library for his MSc Project in the summer.
- (May) Congratulations to Zack Miller, who received the best poster award for the RIT Computer Science MSc Poster Presentation Session for Spring 2014 (poster).
- (May) Congratulations to Zack Miller, Nidhin Pattaniyil and Christopher Sasarak, who have all successfully completed their MS Projects (their presentation posters are available from the "Publications" link above).
- (Apr.) Kenny Davila's paper describing a state-of-the-art classifier for handwritten mathematical symbols has been accepted for oral presentation at ICFHR 2014 being held in Greece this coming September.
- (Apr.) Keita Del Valle's paper describing a user study of the min math entry and search interface has been accepted for publication at ACM SIGIR 2014. SIGIR is being held in the Gold Coast, Australia in July.
- (Apr.) Matthias Reichenbach's paper on the effect of rendering math expression in search hits has been accepted for publication at ACM SIGIR 2014.
- (Mar.) The lab has released a number of systems for math recognition and retrieval as open source on GitHub (DPRL GitHub Page). We hope to add some more information to these pages soon.
- (Mar.) Kenny Davila has released his collection of isolated handwritten math symbol classifiers under GNU GPL v.3 (available here). In benchmarking experiments, the best classifiers (using an SVM with RBF kernel) have recognition rates that match or exceed the best published rates for the MathBrush handwritten symbol data set.
- (Mar.) As part of his upcoming sabbatical, Prof. Zanibbi will be a Visiting Professor at IRCCyN/IVC (Nantes, France) during June and July, and at the University of Waterloo Computer Science Department (Canada) in October.