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

Abstract

Crime generates significant losses, both human and economic. Every year, billions of dollars are lost due to attacks, crimes, and scams. Surveillance video camera networks generate vast amounts of data, and the surveillance staff cannot process all the information in real-time. Human sight has critical limitations. Among those limitations, visual focus is one of the most critical when dealing with surveillance. For example, in a surveillance room, a crime can occur in a different screen segment or on a distinct monitor, and the surveillance staff may overlook it. Our proposal focuses on shoplifting crimes by analyzing situations that an average person will consider as typical conditions, but may eventually lead to a crime. While other approaches identify the crime itself, we instead model suspicious behavior—the one that may occur before the build-up phase of a crime—by detecting precise segments of a video with a high probability of containing a shoplifting crime. By doing so, we provide the staff with more opportunities to act and prevent crime. We implemented a 3DCNN model as a video feature extractor and tested its performance on a dataset composed of daily action and shoplifting samples. The results are encouraging as the model correctly classifies suspicious behavior in most of the scenarios where it was tested. For example, when classifying suspicious behavior, the best model generated in this work obtains precision and recall values of 0.8571 and 1 in one of the test scenarios, respectively.

An expert video surveillance system to identify and mitigate shoplifting in megastores

“…Gated Recurrent Networks (GRU) [19] is another type of recurrent neural network typically used in sequence prediction problems. Both LSTM and GRU aim to keep long-term dependencies effectively while handling the vanishing gradient problems.…”

“…Theft by customers has out-turned many direct and indirect effects on the retailer's income, causing massive losses to the business. The persons involved in this kind of illegal act are called shoplifters, and activity is called shoplifting [19,29]. Since no practical methodology is yet available to identify shoplifting, we intend to create an intelligent and automated indoor surveillance system for stores/shops, to be able to capture these shoplifters by identifying their stealing actions effectively.…”

Using Social Signals to Predict Shoplifting: A Transparent Approach to a Sensitive Activity Analysis Problem

“…The development of an automated system which can detect individuals who exhibit the behaviours associated with shoplifting and draw them to the attention of security staff could help to alleviate some of these challenges. The work conducted by Mascorro et al [6] aimed to detect suspicious behaviour using a 3D CNN model, proposing a new method for processing surveillance footage by segmenting each video into three distinct sections:…”

“…These results are encouraging as they demonstrate that an individual's behaviour can be a strong indicator for predicting whether they are a potential criminal. However, while deep learning models [6,7], have been shown to achieve state of the art results for many computer vision problems, such models function effectively as a black box, as shown in Figure 1, making it almost impossible to determine how the algorithm came to its decision [8].…”

“…Figure 1. The black box model for shoplifting detection, where a raw video sequence is used to train a black box algorithm (such as a 3D CNN as in [6]) to detect suspicious individuals. The nature of these models makes them difficult to interpret and susceptible to bias in the training data.…”