I have a video which contains of cars, a junction and people.
the quality is quite low and the people are very small as its a birds eye camera
Now i have 10000 frames.. i have labelled manually 45000 of them..
1 abnormal in a mat file.
its basically 45000 rows with a -1 or 1 inside.
Then i load my video and convert to frames.. carry out some analysis, optical flow etc and create feature vector.
now... how can i use the 45000 frames that i have labelled in my mat file together with my features of these 45000 frames to test the other remaining frames to see if they will be -1 or 1??
i am new at matlab so any help would be great... thanks!
also i tried to use adaboost but i think i wrote the code wrong...
basically i want to detect abnormalities in my remaining frames based on the labelling of the previous.
please let me know if i should label them in another way and how i can train/test...
the dataset i am using is
(http://www.eecs.qmul.ac.uk/~ccloy/downloads_qmul_junction.html) maybe if you check this out you can get an idea of my data! thanks