Given an input video sequence of one person who conducted a sequence of continuous actions

we consider the problem of jointly segmenting and recognizing actions.To recognize the activities in videos

we propose a discriminative approach to this problem within a semi-Markov model framework

where we are able to define a set of features over input-output space that captures the characteristics on boundary frames

action segments and neighboring action segments

respectively.A Viterbi-like algorithm is devised to help efficiently solve the induced optimization problem.Experiments on a variety of datasets demonstrate the effectiveness of the proposed method.