An apparatus for recognizing contents of a video made of picture frames includes a splitting unit that splits the picture frames into a plurality of sets of video shots based on cut points indicating a change of screen; a similar-video-shot extracting unit that extracts video shots similar to each of the video shots from among the sets of video shots; a maximum-count-video-shot extracting unit that counts a number of similar video shots for each of the video shots and extracts a maximum count video shot having a maximum count of the similar video shots; and a representative-video-shot determining unit that takes the maximum count video shot as a representative video shot for the video.