Manmatha, R.

Profile Picture
Email Address
Birth Date
Research Projects
Organizational Units
Job Title
Research Associate Professor, Department of Computer Science
Last Name
First Name
Computer Sciences
Computer vision
Digital libraries
Document image analysis
Image and video retrieval
Information retrieval
Multimedia indexing and retrieval
Professor Manmatha is broadly interested in the areas of information retrieval, computer vision and document image processing. His recent work focuses on statistical approaches from information retrieval and machine learning for automatically annotating and retrieving images and videos. He also works on the recognizing and retrieving handwritten manuscripts. He and his students have built the first demonstration system for automatically retrieving handwritten historical manuscripts (George Washington's manuscripts). He has also worked on meta-search in information retrieval, detecting text in images and image matching.
Prof. Manmatha was area co-chair for Audio, Video and Image Retrieval for the ACM SIGIR conference in 2001, 2003, 2004 and 2005. He has served on program committees for a number of conferences including CIKM, CIVR, CVPR, DAS, DIAL, ICDAR, ICIP, WACV. He is an associate editor for Pattern Recognition Letters and was previously associate editor for ACM TOIS. He has also conducted tutorials on image retrieval and co-chaired and organized workshops in image retrieval and document recognition. Prof. Manmatha is a member of ACM and IEEE.

Search Results

Now showing 1 - 1 of 1
  • Publication
    Using Maximum Entropy for Automatic Image Annotation
    (2004-01-01) Jeon, Jiwoon; Manmatha, R
    In this paper, we propose the use of the Maximum Entropy approach for the task of automatic image annotation. Given labeled training data, Maximum Entropy is a statistical technique which allows one to predict the probability of a label given test data. The techniques allow for relationships between features to be effectively captured. and has been successfully applied to a number of language tasks including machine translation. In our case, we view the image annotation task as one where a training data set of images labeled with keywords is provided and we need to automatically label the test images with keywords. To do this, we first represent the image using a language of visterms and then predict the probability of seeing an English word given the set of visterms forming the image. Maximum Entropy allows us to compute the probability and in addition allows for the relationships between visterms to be incorporated. The experimental results show that Maximum Entropy outperforms one of the classical translation models that has been applied to this task and the Cross Media Relevance Model. Since the Maximum Entropy model allows for the use of a large number of predicates to possibly increase performance even further, Maximum Entropy model is a promising model for the task of automatic image annotation.