| Recently still image-based human action recognition has become an active research topic in computer vision and pattern recognition. It focuses on identifying a person's action or behavior from a single image. Unlike the traditional action recognition approaches where videos or image sequences are used, a still image contains no temporal information for action characterization. It is more challenging to perform still image-based action recognition than the video-based, given the limited source of information as well as the cluttered background for images collected from the Internet.;Based on the emerging research in recent years, we dig into all possible cues from a single image: the whole image, human figure, action objects, and human-object interaction. We try to use supervised/semi supervised learning by using random forest/support vector machine. For the object labels, we also try automatic localization or manually label all the training/testing images. We have proved that besides the action scene and human action contour, action objects and their relations with centered human figure is important to still image human action recognition. |
