A Revisit of Sparse Coding Based Anomaly Detection in Stacked RNN Framework

Luo, Weixin; Li, Wen; Gao, Shenghua

doi:10.1109/iccv.2017.45

Cited by 708 publications

(573 citation statements)

References 16 publications

Supporting

Mentioning

570

Contrasting

Order By: Relevance

“…Zhao et al [44] proposed to use 3D convolution based reconstruction and prediction. Luo et al [24] iteratively update the sparse coefficients via a stacked RNN to detect anomalies in videos. Liu et al [22] train a frame prediction network by incorporating different techniques including gradient loss, optical flow, and adversarial training.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection

Gong

Liu

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

1,235

686

View full text Add to dashboard Cite

Deep autoencoder has been extensively used for anomaly detection. Training on the normal data, the autoencoder is expected to produce higher reconstruction error for the abnormal inputs than the normal ones, which is adopted as a criterion for identifying anomalies. However, this assumption does not always hold in practice. It has been observed that sometimes the autoencoder "generalizes" so well that it can also reconstruct anomalies well, leading to the miss detection of anomalies. To mitigate this drawback for autoencoder based anomaly detector, we propose to augment the autoencoder with a memory module and develop an improved autoencoder called memory-augmented autoencoder, i.e. MemAE. Given an input, MemAE firstly obtains the encoding from the encoder and then uses it as a query to retrieve the most relevant memory items for reconstruction. At the training stage, the memory contents are updated and are encouraged to represent the prototypical elements of the normal data. At the test stage, the learned memory will be fixed, and the reconstruction is obtained from a few selected memory records of the normal data. The reconstruction will thus tend to be close to a normal sample. Thus the reconstructed errors on anomalies will be strengthened for anomaly detection. MemAE is free of assumptions on the data type and thus general to be applied to different tasks. Experiments on various datasets prove the excellent generalization and high effectiveness of the proposed MemAE.

show abstract

Section: Related Workmentioning

confidence: 99%

“…AE-Conv2D [9] 0.850 0.800 0.609 AE-Conv3D [44] 0.912 0.771 -TSC [24] 0.910 0.806 0.679 StackRNN [24] 0…”

Section: Experiments On Video Anomaly Detectionmentioning

confidence: 99%

Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection

Gong

Liu

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

1,235

686

View full text Add to dashboard Cite

show abstract

“…For verification of the general applicability of our model, we carry out extensive experiments with two types of mainstream action classifiers: a 3D-conv network C3D [59] and a two-stream structure TSN [62]. In addition, we evaluate the proposed approach on 3 different-scale datasets, i.e., UCF-Crime [58], ShanghaiTech [43] and UCSD-Peds [35]. The experimental results demonstrate that our model advances the state-of-the-art performance of weakly supervised anomaly detection.…”

Section: Introductionmentioning

confidence: 99%

Graph Convolutional Label Noise Cleaner: Train a Plug-And-Play Action Classifier for Anomaly Detection

Zhong

Kong

et al. 2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

401

243

View full text Add to dashboard Cite

Video anomaly detection under weak labels is formulated as a typical multiple-instance learning problem in previous works. In this paper, we provide a new perspective, i.e., a supervised learning task under noisy labels. In such a viewpoint, as long as cleaning away label noise, we can directly apply fully supervised action classifiers to weakly supervised anomaly detection, and take maximum advantage of these well-developed classifiers. For this purpose, we devise a graph convolutional network to correct noisy labels. Based upon feature similarity and temporal consistency, our network propagates supervisory signals from high-confidence snippets to low-confidence ones. In this manner, the network is capable of providing cleaned supervision for action classifiers. During the test phase, we only need to obtain snippet-wise predictions from the action classifier without any extra post-processing. Extensive experiments on 3 datasets at different scales with 2 types of action classifiers demonstrate the efficacy of our method. Remarkably, we obtain the frame-level AUC score of 82.12% on UCF-Crime.

show abstract

“…The main drawback is the high computational cost in finding combination coefficients due to sparse representation. Some studies thus attempt to reduce the complexity by modifying the learning algorithms and/or data structures [26,28]. Beside window-based split, 3D patches are also determined using keypoint detectors [5] while other researchers attempt to learn the relation between training patches according to their distribution [30] or graph-based representation [20].…”

Section: Sparse Codingmentioning

confidence: 99%

“…Ped2 Conv-AE [11] 0.702 0.900 Discriminative learning [7] 0.783 -Hashing filters [54] -0.910 Unmask late fusion [16] 0.806 0.822 AMDN (double fusion) [51] -0.908 ConvLSTM-AE [27] 0.770 0.881 DeepAppearance [42] 0.846 -FRCN action [14] -0.922 TSC [28] 0.806 0.910 Stacked RNN [28] 0.817 0.922 AbnormalGAN [37] -0.935 GrowingGas [46] -0.941 Future frame prediction [25] 0.851 0.954 Our proposed method 0.869 0.962 Table 1. Frame-level performance (AUC) of anomaly detection on the CUHK Avenue and UCSD Ped2 datasets.…”

Section: Avenuementioning

confidence: 99%

Anomaly Detection in Video Sequence With Appearance-Motion Correspondence

Nguyen

Meunier²

2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

355

148

View full text Add to dashboard Cite

Anomaly detection in surveillance videos is currently a challenge because of the diversity of possible events. We propose a deep convolutional neural network (CNN) that addresses this problem by learning a correspondence between common object appearances (e.g. pedestrian, background, tree, etc.) and their associated motions. Our model is designed as a combination of a reconstruction network and an image translation model that share the same encoder. The former sub-network determines the most significant structures that appear in video frames and the latter one attempts to associate motion templates to such structures. The training stage is performed using only videos of normal events and the model is then capable to estimate frame-level scores for an unknown input. The experiments on 6 benchmark datasets demonstrate the competitive performance of the proposed approach with respect to state-ofthe-art methods. AbstractThis supplementary material provides these contents:• ROC curves of our frame-level scores on the CUHK Avenue and UCSD Ped2 datasets, and Precision-Recall (PR) curves on the traffic datasets.

show abstract

A Revisit of Sparse Coding Based Anomaly Detection in Stacked RNN Framework

Cited by 708 publications

References 16 publications

Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection

Memorizing Normality to Detect Anomaly: Memory-Augmented Deep Autoencoder for Unsupervised Anomaly Detection

Graph Convolutional Label Noise Cleaner: Train a Plug-And-Play Action Classifier for Anomaly Detection

Anomaly Detection in Video Sequence With Appearance-Motion Correspondence

Contact Info

Product

Resources

About