ankitvora7 / gesture-recognition-hmm Goto Github PK

Dear Mr.Ankit Vora,

Could you please send me a hint or paper to understand the code or how it works?or any paper explains the concept.
thanks in advance

Best regards,
Kerolos Ghobrial

Dear Mr.Ankit Vora,

I have a huge skeleton dataset. it is 27 activities performed by 8 subjects and 4 times. "http://www.utdallas.edu/~cxc123730/UTD-MHAD.html". first of all, i trained i used the kmeans for extracting the cluster center for the whole training data which is 721 activities. after that I computer forward-backword algorithms and save the parameters "Observations"a, b ,c , cluster centers. i only modified the number of states to 54 and number of the cluster to 265.
when I tested it I did not obtain a low accuracy which is 5%.

• I want to know how can i tuning the paramenters # states and # of clusters?
• can i use GMM instead of K-means?
• I want to use this method for classifying online . my idea instead of training all the data together. every activity will be trained individually. what is your opinion ? i should use anther Classifiers like Maximum Entropy Markov Model "MEMM"or HMM is enough.?

thanks in advanced :)