Short pdf version
Prof. Mohan M. Trivedi,
Department of Electrical and Computer Engineering, University of
California, San Diego.
Machine Vision, Machine Learning, Signal
Processing, Sensor Fusion and Multi-Modal/Multi-Sensory Data
Engineering, University of California, San Diego (UCSD), Expected
Computer Engineering, University of California, San Diego (UCSD), awarded,
Electrical Engineering, Indian Institute of Technology, Bombay (IITB),
Skills and Tools:
C, C++, Python, Perl,
OpenCV, MATLAB, Qt,
OpenGL, ADTF, HTK, Microsoft Visual Studio, Boost C++ Libraries
Makefile, Boost build
IP/Network Cameras, Microphone array, Familiar with
Android-based devices (smart phones and Google-Glass)
vision, machine learning, signal processing, speech processing,
multimodal fusion, audio-visual expression recognition, visual saliency
modeling, Distributed and Synchronized
Multimodal-Multisensory testbeds design (smart spaces and smart cars).
Machine, Adaboost, Random Forest, Clustering (k-means, spectral),
Guassian Mixture Model, Hidden Markov Models, Probabilistic modeling,
Random Processes(A+), Parameter Estimation I(A),
Statistical Learning II(A), Speech Compression(A), Computer Vision and
Multimodal System(A+), Computer Vision II(A+), Speech Recognition(A),
Numerical Optimization (A+), Introduction to Computer Graphics1
Adaptive Signal Processing(AA), Wireless and Mobile Communication(AA),
Communication Systems Theory(AA), Advanced Programming in C++(AA),
Wavelets2(AU), Computer Organization, Architecture and
Microprocessors(AB), Control Systems(AA), Signals and Systems(AA),
Digital Circuits(AB), Computational Methods(AA).
2 audited - Extra courses taken
out of interest)
Qualcomm India Pvt. Ltd.
Jul 06 - Jul 08
the post-processing features like Equalizer/Filters, Re-sampler,
Automatic Dynamic Range Control, Doppler Effect, audio visualization
effects like Spectrum Analyzer.
developed and implemented dynamic priority based memory allocation (malloc)
to enhance the concurrent playback of different decoders. The
internal memory is a rare and useful resource and a bottle-neck for
the concurrency requirement. Malloc not only uses the resource
optimally but also prioritizes the decoders to facilitate the memory
requirement for higher priority decoders.
Awarded special recognition (QUALSTAR)
for the contribution in the project.
Intern, QCT Multimedia
RnD & Standard, Qualcomm Incorporated, San Diego, CA
10 - Sept 10
Improved and Implemented high-quality low delay music
coding scheme for next generation vocoders to handle music signal
over voice channel at low bit rates (24/32 Kbps). Project involved
coding of LPC residual in MDCT domain. Developed and implemented an
innovative idea of identifying regions of interest and coding them
in order of their perceptual importance.
Awarded QUALSTAR for the exceptional contribution.
Contribution was finally led to a patent publication.
IEEE Intelligent Transportation Systems
Intelligent Vehicle Symposium
Feb 12 -
Feb 12 - Present
Electrical and Computer Engineering Department, San Diego, CA
Dec 10 -
Laboratory for Intelligent and Safe Automobiles
(LISA): core group member involved in design, development and
operation of powerful and novel real-world driving testbeds for
research in intelligent vehicle systems. Current research includes
developing, data driven, machine-vision and machine-learning based
methods to analyze driver behavior and intent along with surround
situation criticality (object detection) for improving Active Safety
Systems. Proposed a distributed camera framework to robustly and
continuously monitor driver's state, which improved system's
operational range over two-times. Developed a probabilistic model to
infer driver's focus-of-attention. Deployed and demonstrated a
real-time system, 'Attention-Guard', in Audi-A8 concept car (Audi
sponsored project). The system is expected to enter pre-production
stage for further testing and future commercial deployment (media
and press releases:
CNN (page 3)).
- Analyzed a big
naturalistic driving data (>100 hours, collected over 2+ years) for
novel research and algorithm development.
- Designed and developed a
time-synchronized multi-sensory data visualization interface
(multiple videos, vehicle dynamics (time-series) data, lidar/radar tracked objects, GPS map) to identify trends and
relationships between different pieces of data.
'Semiautomatic-Intelligent-Annotator', a platform for quick, simple
and accurate data annotation - an essential step for supervised
Mentored junior graduate
and undergraduate students.
Closely collaborated with
industry sponsors (Electronic Research Laboratory and Audi,
aspect involves: SIFT, linear regression and circular-Hough
transform (facial feature and iris detection/tracking); least square
estimation (3D head pose calculation); saliency modeling using
object detectors (car and pedestrian detection); Random Forest
classifier and domain knowledge (human intent and focus-of-attention
Dec 08 - Dec 10
Computer Vision and Robotics Research Laboratory (CVRR):
designed and developed distributed and synchronized multimodal,
multisensory smart spaces and intelligent vehicles testbeds for
understanding human/driver behavior. Analyzed and developed audiovisual
methods to recognize emotional content of the video data. Completed an
analysis to the most important audio cues to infer the emotion based on
machine learning approach and extended the analysis to make it robust to
background noise, e.g. in driving scenario. Furthermore, a novel facial
expression recognition system, utilizing cross modal (audio cues) data
association, is developed to address the effect of speech generation on
spontaneous facial expression, and demonstrated improved recognition
rate. Relevant publications are listed below.
analysis involves: silence, voiced/unvoiced region identification; pitch
and energy contour calculation (prosodic feature extraction); MFCC
(spectral feature extraction); Support Vector Machine (SVM) classifier
analysis involves: image registration techniques (face alignment); Gabor
wavelets and LBP (low level feature extraction); Adaboost and
Correlation-based Feature Subset selection (feature selection); SVM
classifier (facial expression recognition)
Sept 09 - Nov 09
Developed and implemented an expression recognition
system using face images. Project involved face detection, feature
extraction (Gabor wavelets) and selection (using Adaboost), and
classification module (SVMs) implementation. Further, relaxed
frontal pose constraints on face images and used image registration
technique as a front end to the system, and demonstrated improved
Instructor: Prof Nuno Vasconcelos.
Apr 09 -
and implemented a psycho-acoustical model of auditory perception as
front-end to Hidden Markov Model (HMM) based ASR system. Addressed
robustness of the model over conventional MFCC based ASR system
under different noisy conditions.
Instructor: Prof Bhaskar Rao.
IITB Electrical Engineering Department,
Jul 05 - May 06
the optimization problem with minimum energy and time as constraint.
Proved that the optimal solution falls in the category of NP-hard
problem. Proposed a suboptimal solution for the problem.
Tawari, Andreas Møgelmose, Sujitha Martin, Thomas Moeslund and Mohan M. Trivedi, "First Person Attention Estimation by Simultaneous Analysis
of Viewer and View,"
Transportation Systems Conference,
Ashish Tawari, Kuo Chen
and Mohan M. Trivedi, "Where is the driver looking: Analysis of
Head, Eye and Iris for Robust Gaze Zone Estimation,"
Intelligent Transportation Systems Conference,
Eshed Ohn-Bar, Sujitha
Martin, Ashish Tawari and Mohan Trivedi, "Title Towards
Understanding Driver Activities from Head and Hand Coordinated
IEEE international Conference on
Pattern Recognition (ICPR), 2014. [Nominated for best
industry related paper award]
Ashish Tawari and Mohan M.
Trivedi, "Robust and Continuous
Estimation of Driver Gaze Zone by Dynamic
Analysis of Multiple Face Videos",
IEEE Intelligent Vehicles Symposium, June 2014.
Ashish Tawari, Sayanan
Sivaraman, Mohan M. Trivedi, Trevor Shannon and Mario Tippelhofer, "Looking-in
and Looking-out Vision for Urban Intelligent Assistance: Estimation
of driver Attentive State and Dynamic Surround for Safe Merging and
IEEE Intelligent Vehicles Sym, June 2014.
Tawari, Sujitha Martin and Mohan M. Trivedi, "Continuous Head
Movement Estimator (CoHMET) for Driver Assistance: Issues,
Algorithms and On-Road Evaluations," IEEE Transactions on
Intelligent Transportation Systems, 2014.
Martin, Ashish Tawari and Mohan M. Trivedi,
"Towards Privacy Protecting Safety Systems for Naturalistic Driving
Videos," IEEE Transactions on Intelligent Transportation Systems,
Ashish Tawari, Sujitha Martin, and Mohan M. Trivedi, "Surveillance
for Safety Critical Events: In-Vehicle Video Networks for Predictive
Driver Assistance Systems," accepted and to
appear in Computer Vision and Image
Understanding (CVIU), 2014.
Tawari and Mohan M. Trivedi, "Face Expression Recognition by Cross
Modal Data Association," IEEE Transactions on Multimedia,
vol.15, no.7, pp. 1543-1552, Nov. 2013.
Tawari and Mohan M. Trivedi, “Head Dynamic Analysis: A Multi-view
Framework,” New Trends in Image Analysis and Processing, IEEE
Int. Conf. on Image Analysis and Processing , September 2013.
Tawari and Mohan M. Trivedi, "Audio-Visual Data Association for Face
Expression Analysis," 21st IEEE International Conference on
Pattern Recognition (ICPR), November, 2012. [oral
Yamada, Ashish Tawari and Mohan M. Trivedi, "In-Vehicle Speaker
Recognition Using Independent Vector Analysis," 15th IEEE
Intelligent Transportation Systems Conference, (ITSC),
Tawari and Mohan M. Trivedi, "Audio Visual Cues in
Driver Affective State Characterization: Issues and Challenges in
Developing Robust Approaches", IEEE Int. Joint Conf. on Neural
Networks, July 2011. [oral
Ashish Tawari and Mohan
M. Trivedi, "Speech
Emotion Analysis in Noisy Real-World Environment," 20th
International Conference on Pattern Recognition (ICPR),
Tawari and Mohan M. Trivedi, "Speech Emotion Analysis: Exploring the
Role of Context," IEEE Transactions on Multimedia,
vol.12, no.6, pp. 502-509, Oct. 2010.
Certificate of Appreciation for Outstanding contributions to UC San
Diego community, 2013.
QUALSTAR, a reward program for exceptional contribution, for
improving the lower band MDCT encoding performance in the design of
4th generation vocoder to handle music signal over voice
channel (Sept 2010).
Honorable Mention, 2010 IEEE Intelligent Vehicles Symposium Ph.D.
Forum (Jun 2010).
Recipient of Powell fellowship 2008-2011 at UCSD.
QUALSTAR for designing and implementing MALLOC feature (April 2008).
QUALSTAR for providing innovative ideas and technical assistance in
development of the Low Power MP3 experiment
QUALSTAR, for contribution in evaluating Sony VME technology for
audio subsystem (Dec 2007).
of Hostel Organizational Color, 2004.
Recipient of Certificate of Excellence from EESA for poster
presentation in "Image processing"
Recipient of Hostel
Sports Color, 2002-03.
Accomplished a position in top 0.1% out of 125,000 candidates in IIT
Committee, IEEE Intelligent Vehicles Symposium, June 2010.
1st position in inter hostel football match, 2004.
post of Sports Secretary of hostel 5, 2003-04.
in workshops, Techfest, annual technology festival of IITB,
Lead the hostel Kho-Kho
team as captain, 2003.
post of Head Boy of the school, 2000-01.