Learning in addition to Recognizing Activities in Streams of Video Dinesh Govindaraju Activ

Learning in addition to Recognizing Activities in Streams of Video Dinesh Govindaraju Activ www.phwiki.com

Learning in addition to Recognizing Activities in Streams of Video Dinesh Govindaraju Activ

Geyer, Katherine, Contributing Writer has reference to this Academic Journal, PHwiki organized this Journal Learning in addition to Recognizing Activities in Streams of Video Dinesh Govindaraju Activity recognition from video as long as higher functionality Who is presenting agenda item Attendee interest levels Motivation Want it to be automatic in addition to not involve h in addition to generation of models Impractical in the case of many activities Less versatile as you might be constrained to particular aspects of the problem Motivation

Rosemont College PA www.phwiki.com

This Particular University is Related to this Particular Journal

Video Data Observations are extracted movement deltas via face tracking H in addition to label training segments Learn underlying models from training segments Carry out activity recognition Problem Definition Assume underlying models can be approximated by HMMs Use Baum Welch to learn best model using training segments Need to find observation space in addition to number of states Approach – Learning HMMs: Approach – Learning

To find observation space: Run through all training segments in addition to add observations For new observation when doing recognition, augment learned observation matrices Approach – Learning To find number of states, Q ( as long as each activity): Set upper bound as length of longest training segment Iterate over values in addition to generate most likely model using Baum Welch Approach – Learning To find number of states, Q ( as long as each activity): Choose best Q using N-fold cross validation using criterion of discriminative power With best Q, run Baum Welch using a number of sets of r in addition to omly initialized parameters to get a Approach – Learning

Define a window width, w From the beginning, sequentially consider windows of observations (where L is length of entire sequence) Approach – Recognition Calculate likelihood of each window segment Approach – Recognition L Rabinier, A Tutorial on Hidden Markov Models in addition to Selected Applications in Speech Recognition, Proceedings IEEE, 1989 Label middle frame in each window with activity with highest likelihood Approach – Recognition

Activities being observed: Evaluation in addition to Results Observation stream obtained from 87 second long image sequence 1296 individual frames Example frames after face detection: Evaluation in addition to Results Observation sequence first h in addition to labeled Segments showing same activity extracted 4 training segments used to learn each activity Evaluation in addition to Results

Evaluation in addition to Results Once underlying models were learned, calculate likelihood using sliding window Value of 21 was used as long as the window width, w, as this was the average length of training segments Evaluation in addition to Results Evaluation in addition to Results

Carry out recognition using the likelihoods by assigning activities to the frames Compare against h in addition to assigned labels Accuracy approximately 76% Evaluation in addition to Results Algorithm assigned: Evaluation in addition to Results H in addition to assigned: Evaluation in addition to Results

Learn underlying model generating sequence of activities themselves St in addition to ardize lengths of training segments using Dynamic Time Warping in addition to use that as the window width Future Work The End Questions

Geyer, Katherine Signal, The Contributing Writer www.phwiki.com

Geyer, Katherine Contributing Writer

Geyer, Katherine is from United States and they belong to Signal, The and they are from  Santa Clarita, United States got related to this Particular Journal. and Geyer, Katherine deal with the subjects like City/Metropolitan News; Local News

Journal Ratings by Rosemont College

This Particular Journal got reviewed and rated by Rosemont College and short form of this particular Institution is PA and gave this Journal an Excellent Rating.