Are pre-trained CNNs good feature extractors for anomaly detection in surveillance videos?

TL;DR

This paper evaluates the effectiveness of features extracted from pre-trained CNNs for anomaly detection in surveillance videos, highlighting the importance of normalization and achieving competitive results on standard datasets.

Contribution

It provides a comparative analysis of features from four state-of-the-art CNNs and examines the impact of normalization techniques on anomaly detection performance.

Findings

Normalization significantly improves detection accuracy.

Features from CNNs can match state-of-the-art methods on Ped2.

Appearance-based features can be combined with motion analysis for better results.

Abstract

Recently, several techniques have been explored to detect unusual behaviour in surveillance videos. Nevertheless, few studies leverage features from pre-trained CNNs and none of then present a comparison of features generate by different models. Motivated by this gap, we compare features extracted by four state-of-the-art image classification networks as a way of describing patches from security video frames. We carry out experiments on the Ped1 and Ped2 datasets and analyze the usage of different feature normalization techniques. Our results indicate that choosing the appropriate normalization is crucial to improve the anomaly detection performance when working with CNN features. Also, in the Ped2 dataset our approach was able to obtain results comparable to the ones of several state-of-the-art methods. Lastly, as our method only considers the appearance of each frame, we believe that it can be combined with approaches that focus on motion patterns to further improve performance.