Doubly Smoothed Density Estimation with Application on Miners' Unsafe Act Detection

2025-10-25

AI Summary

This work addresses unsupervised image-based anomaly detection of unsafe miner behaviors in underground mine scenes under fixed-camera surveillance. We propose a dual-domain smoothed density estimation method that jointly applies kernel smoothing in both the value and spatial domains to simultaneously reduce bias and variance; additionally, a grid-point approximation technique is introduced to significantly accelerate inference while preserving estimation accuracy. The method operates without labeled anomalies, automatically localizes anomalous sub-image regions, and integrates seamlessly with a lightweight MobileNet classifier for end-to-end detection. Evaluated on large-scale real-world mine surveillance data, it achieves 99% out-of-sample detection accuracy and meets near-real-time processing requirements. Our primary contribution is the first application of dual-domain smoothed density modeling coupled with grid-based approximation to unsupervised anomaly localization in industrial settings—achieving an optimal balance among high accuracy, computational efficiency, and practical deployability.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

We study anomaly detection in images under a fixed-camera environment and propose a emph{doubly smoothed} (DS) density estimator that exploits spatial structure to improve estimation accuracy. The DS estimator applies kernel smoothing twice: first over the value domain to obtain location-wise classical nonparametric density (CD) estimates, and then over the spatial domain to borrow information from neighboring locations. Under appropriate regularity conditions, we show that the DS estimator achieves smaller asymptotic bias, variance, and mean squared error than the CD estimator. To address the increased computational cost of the DS estimator, we introduce a grid point approximation (GPA) technique that reduces the computation cost of inference without sacrificing the estimation accuracy. A rule-of-thumb bandwidth is derived for practical use. Extensive simulations show that GPA-DS achieves the lowest MSE with near real-time speed. In a large-scale case study on underground mine surveillance, GPA-DS enables remarkable sub-image extraction of anomalous regions after which a lightweight MobileNet classifier achieves $approx$99% out-of-sample accuracy for unsafe act detection.

Problem

• Improving anomaly detection accuracy in fixed-camera images
• Developing computationally efficient density estimation method
• Enabling real-time unsafe act detection in mining surveillance

Innovation

• Doubly smoothed density estimator with spatial structure
• Grid point approximation reduces computational cost
• Lightweight MobileNet classifier for anomaly detection