Multifold Classification for Human Action Recognition

Identification of human actions from the video samples is suffered from various challenges like cluttered background, brightness and motion variation. This paper uses the multifold classification for this purpose. The UT interaction video data dataset is used to validate the work. The UT interaction dataset consists of 20 sequence videos with the six different interaction categories of the human-human interaction. The input video is segmented into the frames using a python script. The global features are evaluated with the Histogram of Gradient (HOG) and Histogram of Optical Flow (HOF). The dimension of the HOF and HOG feature vector is reduced by the Principle Component Analysis (PCA) method. The extracted features are fed to the SVM classifier which provids two different classes of six human-human interaction action. Class 1 contains the fighting interaction like Kicking, Punching and Pushing and Class 2 contains the normal interaction like Pointing, Hugging and Hand Shaking. At the second level, motion features Speed up Robust Features (SURF) are extracted. The detected features are represented as the visual words by using the Bag of Visual Word (BoVW) approach. The output of SURF-BoVW features is fed to the Neural Network classifier, which classifies the human-human interaction. The proposed method provided efficient results than the single-level classier like SVM and KNN. The proposed method provides the accuracy upto 99 % for the UT interaction dataset.