Key points are not available for this paper at this time.
In this paper, we have proposed a method to detect abnormal events for human group activities. Our main contribution is to develop a strategy that learns with very few videos by isolating the action and by using supervised learning. First, we subtract the background of each frame by modeling each pixel as a mixture of Gaussians (MoG) to concatenate the higher order learning only on the foreground. Next, features are extracted from each frame using a convolutional neural network (CNN) that is trained to classify between normal and abnormal frames. These feature vectors are fed into long short term memory (LSTM) network to learn the long-term dependencies between frames. The LSTM is also trained to classify abnormal frames, while extracting the temporal features of the frames. Finally, we classify the frames as abnormal or normal depending on the output of a linear SVM, whose input are the features computed by the LSTM.
Kothapalli et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: