******************** denseDT3 ************************
Building a codebook from the huge matrix of dense trajectory features.
- compile as :
g++ -L/opt/lib -pipe -Wall -O3 -ggdb denseDT3.cpp -lopencv_core -lopencv_highgui -lopencv_video -lopencv_imgproc -lavformat -lavdevice -lavutil -lavcodec -lswscale

- run as :
./a.out /Pulsar3/himangi.s/Fall_detection/dense_trajectory_release_v1.2/Hollywood2/ClipSets/AnswerPhone_train.txt Codebook  

-Format of /Pulsar3/himangi.s/Fall_detection/dense_trajectory_release_v1.2/Hollywood2/ClipSets/AnswerPhone_train.txt :
video_name1 class(+1/-1)
video_name2 class(+1/-1)
video_name3 class(+1/-1)
...
...
...
video_namek class(+1/-1)

- hardcoded parameters
* location of DenseTrack compiled;
* feature classes;
* video types(avi)


******************** denseDT8 ************************
Feature quantisation
- compile as :
g++ -L/opt/lib -pipe -Wall -O3 -ggdb denseDT8.cpp -lopencv_core -lopencv_highgui -lopencv_video -lopencv_imgproc -lavformat -lavdevice -lavutil -lavcodec -lswscale

- run as :
./a.out /Pulsar3/himangi.s/Fall_detection/dense_trajectory_release_v1.2/Hollywood2/ClipSets/AnswerPhone_train.txt /Pulsar3/himangi.s/Fall_detection/dense_trajectory_release_v1.2/Hollywood2/ClipSets/AnswerPhone_test.txt /Pulsar3/himangi.s/Fall_detection/dense_trajectory_release_v1.2/Hollywood2/AVIClips Codebook AnswerPhone_train.csv   AnswerPhone_test.csv avi vocab_size 1

-command line arguments
* training files filename
video_name1 class(+1/-1)
video_name2 class(+1/-1)
video_name3 class(+1/-1)
...
...
...
video_namek class(+1/-1)

*test files filename
video_name1 class(+1/-1)
video_name2 class(+1/-1)
video_name3 class(+1/-1)
...
...
...
video_namek class(+1/-1)

*path to vocab files
*vocab file prefix as passes in second argument of denseDT3.
*csv file to store the train quantised vectors
*csv file to store the test quantised vectors
*a prefix(unused)
*vocab file size to store the experiments with differen vocab sizes(unused) 
*skip - to quickly train and test with a few samples, skips skip number of files before chosing next train or text sample. (Use 1 for complete testing)