Shopformer: Transformer-Based Framework for Detecting Shoplifting via Human Pose

2025-04-28

AI Summary

Retail theft poses a severe challenge, with traditional manual surveillance achieving merely ~2% detection rates. Existing AI-based approaches rely on pixel-level video analysis, suffering from privacy violations, environmental sensitivity, and high computational overhead. This paper proposes a lightweight, privacy-preserving shoplifting detection framework that abandons raw video inputs and instead models temporal human pose sequences. Our key contributions are: (1) the first shoplifting detection paradigm that directly treats pose sequences as tokens fed into a Transformer architecture; and (2) a customized tokenization strategy specifically designed to capture dynamic pose characteristics. Evaluated on real-world pose data, our method significantly outperforms state-of-the-art temporal anomaly detection models. It enables real-time inference and edge deployment, achieving an optimal trade-off among high accuracy, low power consumption, and strong privacy preservation.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Privacy
• Philosophy and Ethics of AI: Privacy & Security

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems

Abstract

Shoplifting remains a costly issue for the retail sector, but traditional surveillance systems, which are mostly based on human monitoring, are still largely ineffective, with only about 2% of shoplifters being arrested. Existing AI-based approaches rely on pixel-level video analysis which raises privacy concerns, is sensitive to environmental variations, and demands significant computational resources. To address these limitations, we introduce Shopformer, a novel transformer-based model that detects shoplifting by analyzing pose sequences rather than raw video. We propose a custom tokenization strategy that converts pose sequences into compact embeddings for efficient transformer processing. To the best of our knowledge, this is the first pose-sequence-based transformer model for shoplifting detection. Evaluated on real-world pose data, our method outperforms state-of-the-art anomaly detection models, offering a privacy-preserving, and scalable solution for real-time retail surveillance. The code base for this work is available at https://github.com/TeCSAR-UNCC/Shopformer.

Problem

• Detect shoplifting via human pose analysis
• Address privacy and computational limitations of video-based AI
• Provide real-time, scalable retail surveillance solution

Innovation

• Transformer-based model for pose sequence analysis
• Custom tokenization for compact pose embeddings
• Privacy-preserving real-time shoplifting detection