Current research focus is on (a) integration of human and contextual information in analyzing images and video, leading to bio-inspired methods for computer vision; (b) large scale camera networks and associated "big-data" information processing tasks; and (c) bio-medical image informatics. In addition, we continue the work on fundamental problems of segmentation, registration and tracking, and large scale image/video indexing and search, with emphasis on developing robust and scalable methods.

The Bisque image informatics infrastructure, developed by the VRL group and the Center for Bio-Image Informatics is now integrated into the iPlant cyberinfrastructure. Bisque is currently being used in many major laboratories around the world and we are in the process of adding image feature services and indexing services to facilitate large scale image data management.

View Google Scholar Citations for the VRL publications.

Recent Papers

Camera Networks

Object-Based Retrieval

Video Indexing and Tracking

Image and Video Segmentation

Bio-Image Informatics

Malware

Managing Multimedia Databases, Multimedia Database Mining

Research

High resolution imaging of molecules and cells will be critical for understanding complex systems such as the nervous system, whether it be for the localization of specific neuron types within a region of the central nervous system, the branching pattern of dendritic trees, or the localization of molecules at the subcellular levels. In collaboration with researchers in biology, we are developing new image analysis methods for quantifying complex biological phenomena. For more information about the Center for Bio-Image Informatics and the Bisque open-source system, click on the link above.

Humans are adept at complex visual tasks such as detection, tracking and recognition. The first aim of this project is to understand humans visual attention mechanism and the factors influencing it. In particular we investigate the importance of scene context and camera focus in attention modeling. We have developed human attention inspired models to build semantic priors for text detection which outperforms state-of-the-art. Current work focuses on integrating fMRI and eye-tracking data, towards developing robust recognition methods.

Robust methods for segmentation and tracking are critical for quantitative biology. We are developing methods for accurate, automated and interactive 2D/3D segmentation of cellular and sub-cellular structures, with emphasis on high-throughput image analysis. The three main issues that make analyzing microscopic imagery challenging are modeling of cell appearance which varies from one dataset to another, poor signal to noise ratio/clutter, and the large scale nature of the datasets. We have developed interactive static image segmentation methods using Markov Random Fields (MRFs) that utilize a human in the loop to achieve accurate segmentation. Further, we have proposed an approach for parameter adaptation that induces generic high-level priors on topology, facilitating robust tracing of neuronal structures. Recent work focuses on modeling edge appearance and scaling tracing techniques to hundreds of structures. Finally, we have also focused on the problem of detecting deformable structures such as synaptic junctions in Electron Micrographs.

SARVAM is a demonstration of a simple and effective technique for visualizing and classifying malware using image processing. Malware binaries are visualized as gray scale images, with the observation that for many malware families, images belonging to the same family "appear" very similar in their layout and texture. The online demo allows users to upload binary executable files and search for similar malware in the SARVAM database.

During the past few years we have set up a wide area camera network, called the SCALLOPSNet (Scalable Large Optical Sensor Network). About 100 stationary cameras covera an expansive area that can be divided into three distinct regions: inside a building, along urban paths, and in a remote natural reserve. Challenging research issues include geo-calibration, multicamera video summarization and anomaly detection, and multi-camera tracking and activity analysis in wide area networks.

Most state-of-the-art object trackers assume good visual, clutter-free imaging conditions for reliable tracking. In real-life scenarios, encoding and transmission artifacts, object occlusions and scene clutter play major impediments for tracking performance.  Humans on the other hand effortlessly track objects in such very challenging conditions. We aim to leverage  human-generated knowledge to new challenging datasets for tracking objects.

In a large multi-camera setup, information deluge is a common issue when an analyst queries the network for relevant information. In this project we develop algorithms which can simulate human perceptual behavior when observing multiple camera views. This algorithm can significantly reduce information overload and help prioritize transmission of critical information over a resource-constrained network.

In a large multi-camera setup, information deluge is a common issue when an analyst queries the network for relevant information. In this project we develop algorithms which can simulate human perceptual behavior when observing multiple camera views. This algorithm can significantly reduce information overload and help prioritize transmission of critical information over a resource-constrained network.

Contact

Vision Research Lab
Department of Electrical and Computer Engineering
University of California
Santa Barbara, CA 93106-9560

Tel: (805) 893 7112
Fax: (805) 893 3262
E-mail: manj [at] ece [dot] ucsb [dot] edu

News

  • 2013-11. Congratulations to Santhosh for receiving the outstanding paper award at the recent ICDSC conference on camera networks. He presented two papers and his tracking paper was among the three selected for the "excellent paper" award.
  • 2013-10. Aruna's paper on spine distribution analysis, published in BMC Bioinformatics, is now a "highly accessed article" with over 850 downloads in about four weeks since publication.
  • 2013-09. Thomas Kuo and Karthiok S. will be presenting at the ICCV 2013 conference. Karthik's paper is on combining eye-tracking analysis with computer vision methods for better detection/recognition. Thomas's work investigates camera calibration using unsynchronized videos.
  • 2013-08. Congratulations to Thomas Kuo for successfully defending his PhD research on 8/21. After graduation Thomas will continue to work at Birdeez, which he co-founded, that develops mobile apps to connect people with nature.
  • 2013-08. Congatulations to Santhosh for his two accepted papers to the ICDSC 2013 conference.
  • 2013-08. Dhilung and Lakshman will be presenting their work on a signal processing based approach to malware analysis at the Annual Computer Security Applications Conference (ACSAC) in December.
  • 2013-08. Jiejun's paper on multilabel learning accepted for publication in IEEE Trans. Multimedia. "Multi-label Learning with Fused Multimodal Bi-relational Graph," Jiejun Xu, Vignesh Jagadeesh and B.S. Manjunath.
  • 2013-08. Vignesh Jagadeesh's Trans. Image Processing Paper accepted!
  • 2013-06. Paper on synapse detection accepted for publiction in Pattern Recognition Letters Special Issue. "Synapse Classification and Localization in Electron Micrographs", Vignesh Jagadeesh, James Anderson, Bryan Jones, Robert Marc, Steven K Fisher and B.S Manjunath.
  • 2013-06. The 2013 IEEE Transactions Multimedia Best Paper award to Moxley, Mei and Manjunath's paper on video annotations through graph mining.
  • 2013-05. Diana Delibaltov will be presenting her segmentation/fusion work at the upcoming MICCAI conference in Nagoya, Japan, in September. Congratulations to Diana!
  • 2013-04. Paper accepted for publication in IEEE Transactions Multimedia. The paper by Jiejun et. al. describes a graph based search, re-identification and retrieval of moving objects in a distributed camera network.
  • 2013-03. VRL graduate Dr. Wei-Ying Ma delivers the ECE 50th Anniversary Distinguished Lecture, "The Next Frontier for Web Search and Knowledge Mining". Dr. Ma received his PhD in 1997 and is currently an Assistant Managing Director at Microsoft Research Asia, Beijing.
  • 2013-03. Congratulations to Jiejun Xu, Carter De Leo and Vignesh Jagadeesh for completing their PhD. Jiejun has joined HRL in Malibu, Carter is with Google and Vignesh has started working at eBay.
  • 2013-01. Two papers accepted to ACM Transactions on Sensor Networks. Thomas Kuo's work is about geo-calbrating fixed camera nodes using mobile devices, and Carter De Leo's paper is about multicamera video summarization.
  • 2012-08. Professors Bill Smith and Manjunath awarded a new NIH R01 5 year project to study morphogenesis of developing embryos.