Apply HMM to Human Activities(HMM-HA)

Mativation

1.To apply HMM to some human activities and try to recognize them.

2.Tempt to deduce more higher level meaning of activies.
Progress
Data or Report
Code

Sept.1-15 2007: Decoding the data and learn Kevin's toolbox for DHMM

    
Sept.16-20 2007: Preliminary Experiment to simulate the activities data  

Sept.21-Oct. 10 2007: Design and program HMM-HA(DHMM) to run on 16 activities from raw data.

Experiment 1(EXP1):
Basic HMM-HA with WALK and RUN Models
DataSet1
1.m
Oct. 11-15 Make a comprehensive comparison of all types of sensor data
Report1
  

Oct.11-24 Carry out more experiments to see how well the HMM can do and verifiy some conclusions.


Experiment 2(EXP23): Testing Bi-Direction Activities
Experiment 3(EXP3): Adding more training data from Exp5 and 6
Experiment 4(EXP4): Adding Silence Model
Experiment 5(EXP5): Recognizing Number of people?

Report2

Report3

DataSet2
2.m
Nov. 20-Dec.4 Using Wavelet to denoise the signal
Report_Denoise_1
3.m
     

Data

The data are from the experiments carried out at U.S. Army Research Laboratory (ARL).

HMM Toolbox

*Kevin Murphy http://www.cs.ubc.ca/~murphyk/Software/HMM/hmm.html
*Matlab Statistics Toolbox- only for DHMM
HTK:http://htk.eng.cam.ac.uk/
GPDSHMM: http://www.gpds.ulpgc.es/download/
NetLab: http://www.robots.ox.ac.uk/~parg/software.html
H2M http://www.tsi.enst.fr/~cappe/h2m/h2m.html

HMM Papers

A tutorial on Hidden Markov Models and selected applications in speech recognition, L. Rabiner, 1989, Proc. IEEE 77(2):257--286.