Ajay Divakaran's talk
bhiksha raj
Ajay Divakaran of Sarnoff will give a seminar in GHC 6501 at 11am on Thursday.
Please try to attend. Abstract attached.
-Bhiksha
----------------------------
SEMINAR: Some applications of Micro and Macro Classification of Audio
Speaker: Ajay Divakaran, (Sarnoff Labs, Fellow IEEE)
Where: GHC 6501
When: 11.00 AM, 8th September 2011, Thursday
Abstract:
We present a suite of audio classification applications ranging from
macro-classification into broad categories such as speech, music,
cheering to micro-classification of gunshots recordings into specific
weapon types. We will briefly review an automatic sports highlights
extraction enabled DVD recorder that detects a mixture of the crowd’s
cheering and the commentator’s excited speech. We will then cover
weapon identification through classification of gunshot recordings.
Gunshot recordings have the potential for both tactical detection and
forensic evaluation particularly to ascertain information about the
type of firearm and ammunition used. Perhaps the most significant
challenge to such an analysis is the effect of recording conditions on
the audio signature of recorded data. We present a first study of
using an exemplar embedding approach to automatically detect and
classify firearm type across different recording conditions. We
demonstrate that a small number of exemplars can span the space of
gunshot audio signatures and that this optimal set can be obtained
using a wrapper function. By projecting a given gunshot to the
subspace spanned by the exemplar set a distance measure/feature vector
is obtained that enables comparisons across recording conditions. We
also investigate the use of a hierarchy of gunshot classifications
that assists in improving finer level classification by pruning out
gunshot labeling that is inconsistent with its higher level type. The
embedding based approach can thus be used both by itself and as a
pruning stage for other search techniques. We then present an approach
to audio classification for event identification in open source video.
We present preliminary results with novel clustering methods that
outperform Gaussian Mixture Model based methods. We identify the
principal challenges posed by audio classification for open source
video and describe avenues for future research. Finally, we describe a
couple of new applications of audio classification.
Bio:
Ajay Divakaran, PhD is a Technical Manager at SRI International
Sarnoff. He has developed several innovative technologies for
multimodal systems for both commercial and government programs over
the past 16 years. He currently leads SRI Sarnoff’s projects on
Modeling and Analysis of Human Behavior for the DARPA SSIM project,
Audio Analysis for Event Detection in Open Source Video in the IARPA
Aladdin program, and People, Vehicle and Vessel tracking for ONR and
JIEDDO-DHS among others. He worked at Mitsubishi Electric Research
Labs for ten years where he was the lead inventor of the world’s first
sports highlights playback enabled DVR, as well as a manager
overseeing a wide variety of product applications of machine learning.
He was elevated to Fellow of the IEEE in 2011 for his contributions to
multimedia content analysis. He developed techniques for recognition
of agitated speech for his work on sports highlights. He established a
sound experimental and theoretical framework for human perception of
action in video sequences, as lead-inventor of the MPEG-7 video
standard motion activity descriptor. He serves on TPC’s of key
multimedia conferences and served as an associate editor of the IEEE
transactions on Multimedia from 2007 to 2011 and has two books and
over 100 publications to his credit as well as 38 issued patents. He
received his Ph.D. degree in Electrical Engineering from Rensselaer
Polytechnic Institute in 1993.