Deep anomaly detection through visual attention in surveillance videos

Abstract

This paper describes a method for learning anomaly behavior in the video by finding an attention region from spatiotemporal information, in contrast to the full-frame learning. In our proposed method, a robust background subtraction (BG) for extracting motion, indicating the location of attention regions is employed. The resulting regions are finally fed into a three-dimensional Convolutional Neural Network (3D CNN). Specifically, by taking advantage of C3D (Convolution 3-dimensional), to completely exploit spatiotemporal relation, a deep convolution network is developed to distinguish normal and anomalous events. Our system is trained and tested against a large-scale UCF-Crime anomaly dataset for validating its effectiveness. This dataset contains 1900 long and untrimmed real-world surveillance videos and splits into 950 anomaly events and 950 normal events, respectively. In total, there are approximately ~ 13 million frames are learned during the training and testing phase. As shown in the experiments section, in terms of accuracy, the proposed visual attention model can obtain 99.25 accuracies. From the industrial application point of view, the extraction of this attention region can assist the security officer on focusing on the corresponding anomaly region, instead of a wider, full-framed inspection.

Efficient Spatio-Temporal Modeling Methods for Real-Time Violence Recognition

“…Contrary to most of the works using the LSTM or 3D CNN for temporal modeling, Jain and Vishwakarma aggregated multiple frames called Dynamic Image (DI) to represent motion features in a single frame and fed it into 2D CNN [66]. Background subtraction was also used in violence detection to reduce the influence of disturbing backgrounds [36]. Su et al used graph convolution for violence detection with an assistance of a pose estimation model [12].…”

“…Kim et al computed the motion features with edge and color orientation histograms to generate a spatio-temporal saliency map [35]. Nasaruddin et al proposed background (BG) subtraction method to blur uninteresting areas in the surveillance video by using the binary bitmaps of each frame [36]. Some works utilized spatial and temporal attention modules in video action recognition to reduce redundant information over space and time [37]- [41].…”

Substraksi Latar Menggunakan Nilai Mean Untuk Klasifikasi Kendaraan Bergerak Berbasis Deep Learning

“…Sebagaimana dijelaskan pada [15], ukuran blok 4x4 mampu menghasilkan informasi tekstur yang paling baik, jika dibandingkan dengan ukuran blok lain, yaitu ukuran 2x2 dan 6x6. Pada perbandingan tersebut, ukuran blok 2x2 menghilangkan terlalu banyak informasi tekstur, sedang blok 6x6 menampilkan detail yang tidak penting pada blok biner yang dihasilkan.…”

Criminal Intention Detection at Early Stages of Shoplifting Cases by Using 3D Convolutional Neural Networks

“…Their model includes a ranking loss function and trains a fully connected neural network for decision-making. In a similar context, Nassarudin et al [29] presented a deep anomaly detection approach. They implemented a bilateral background subtraction, use the pretrained C3D model [24] for feature extraction, and attached a fully connected network to perform regression.…”