Privacy-Preserving, Thermal Vision With Human in the Loop Fall Detection Alert System

Abstract

To support the independent living of older adults in their own homes, it is essential to identify their abnormal behaviors before triggering an automated alert system. Existing normal vision sensing approaches to detect human falls in the activities of daily living (ADL) experienced acceptability issues due to outstanding privacy concerns when they are deployed in personal environments. Besides, false alerts (false-positive) fall detection has not been addressed thoroughly in systems that report abnormal human behaviors as emergency alerts to the information support. This article proposes a novel human-in-the-loop fall detection approach in the ADLs using a low-resolution thermal sensor array. The motivation for enabling a human interactive model, fall detection confirmation, is to influence resource efficiency by reducing false-positive alerts while keeping the false-negative fall predictions as low as possible. The proposed approach is based on the motion sequence classification of human movements using a recurrent neural network. The proposed approach is evaluated with comprehensive experiments using different learning techniques, users, and domestic environment conditions. This article shows a performance accuracy of 99.7% to detect human falls from various typical ADLs.

Enhancing privacy with optical element design for fall detection

“…Additionally, we calculated the average difference (AD) between the normalized original images and the encrypted ones based on L2 norm, to assess the encryption level. For comparison purposes, we included the performance of fall detection based on the original images without encryption, as well as two other hardware encryption methods: the defocus lens [10] and the low-resolution camera [5,6] as well as the software-based compressed sensing method in [7]. Figure 3 illustrates the obtained AP curves for different methods, while Table 1 presents the quantitative AP and AD values.…”

“…In particular, [5] and [6] tackle the privacy concern by leveraging low‐resolution vision sensors. They utilize low‐resolution thermal images as input, which are then processed by either the Three‐Dimensional Convolutional Neural Network [5] or the Recurrent Neural Network [6] for fall detection. Additionally, [7] employs compressed sensing to detect falls while preserving privacy through the use of low‐quality compressed data.…”

“…Recent attention from researchers has been directed towards addressing this issue, leading to the development of various privacy-preserving techniques for computer vision-based fall detection. In particular, [5] and [6] tackle the privacy concern by leveraging low-resolution vision sensors. They utilize low-resolution thermal images as input, which are then processed by either the Three-Dimensional Convolutional Neural Network [5] or the Recurrent Neural Network [6] for fall detection.…”

Exploiting the PIR Sensor Analog Behavior as Thermoreceptor: Movement Direction Classification Based on Spiking Neurons

“…In addition, the use of enhanced resolution IR arrays is contrary to the people's privacy preservation present in the space under study. This is especially important in the people's detection of indoor non-public or personal use areas [12]. In this sense, diverse commercial integrated IR sensor arrays allow us to capture a low-res thermal image, such as the Panasonic ® AMG88xx Series (Grid-Eye ® Infrared Array Sensors) or Omron ® D6T Series MEMS Thermal Sensors.…”