Deep learning: A QRIScloud GPU node experience
Ranju Mandal
About the project
QRIScloud’s special compute service has enabled a Griffith University PhD student to run document image retrieval experiments faster and with greater flexibility.
Ranju Mandal, at Griffith’s Gold Coast School of Information and Communication Technology, Is applying QRIScloud’s GPU (Graphics Processing Unit) node in his experiments. Together with a deep learning technique, his GPU algorithms can retrieve relevant documents by image queries into document repositories.
The goal of Mr Mandal’s research is to make querying and document retrieval based on images (e.g. a handwritten signature, seal or logo) in administrative documents a lot faster, making possible an automated process, rather than manual document scans. For example, a bank or other company could use the application to search its databases for documents featuring a particular signature.
“Deep learning has been used in a wide range of projects involving computer vision, robotics and artificial intelligence. By leveraging powerful GPU-based searching, deep learning is helping these fields to make great strides,” said Mr Mandal, whose own research interests include document image analysis, document image retrieval and pattern recognition.
“The QRIScloud GPU node is fast and provides all the flexibility that a researcher needs.”
Using the Caffe deep learning framework developed by the Berkeley Vision and Learning Center, Mr Mandal plans to use a U.S. University of Maryland public data set involving images from multi-page documents, collected and scanned using a wide variety of equipment over time, with varying resolutions and dimensions.
QCIF eResearch Analyst Cihan Altinay is providing user support for special nodes within QRIScloud, including four large memory and four GPU nodes. Please contact Cihan if you wish to discuss how these special nodes could enhance your research: This email address is being protected from spambots. You need JavaScript enabled to view it..
