Look at Adjacent Frames: Video Anomaly Detection without Offline Training

TL;DR

This paper introduces an online, training-free video anomaly detection method using a randomly-initialized neural network optimized in real-time to identify unusual events based on frequency information shifts between frames.

Contribution

It presents a novel online learning approach with a multilayer perceptron that detects anomalies without offline training, overcoming previous limitations on abnormal frame frequency.

Findings

Achieves strong performance on benchmark datasets.

Operates effectively without offline training.

Detects anomalies based on frequency information shifts.

Abstract

We propose a solution to detect anomalous events in videos without the need to train a model offline. Specifically, our solution is based on a randomly-initialized multilayer perceptron that is optimized online to reconstruct video frames, pixel-by-pixel, from their frequency information. Based on the information shifts between adjacent frames, an incremental learner is used to update parameters of the multilayer perceptron after observing each frame, thus allowing to detect anomalous events along the video stream. Traditional solutions that require no offline training are limited to operating on videos with only a few abnormal frames. Our solution breaks this limit and achieves strong performance on benchmark datasets.