Our propose method focuses to detect and localize anomalous behavior in videos of crowded area means different scenario from dominant pattern. Proposed method consist motion and appearance information therefore different kinds of anomalies can be robustly identified in a wide range of situations. Histogram of oriented gradients can easily captures varying dynamic of crowded environment. Histogram of oriented gradients can also effectively recognize and characterize each frame of each scene. Our method of detecting anomalies using artificial neural network and support vector machine consist both appearance and motion features which extracts this features within spatio temporal domain of moving pixels that ensures robustness to local noise and thus increases accuracy in detection of local anomaly with low computational cost. UCSD dataset which will be used and which consist various situations with varying human crowds as well as traffic data with occlusions when feed to our propose method can achieve significantly higher accuracy probably more for pixel level events detection as compared to any other methods.