Great Expectations: Perceptual Challenges of Visual Surveillance in Lifeguarding

MS-YOLO: A Lightweight and High-Precision YOLO Model for Drowning Detection

A novel detection model, MS-YOLO, is developed in this paper to improve the efficiency of drowning rescue operations. The model is lightweight, high in precision, and applicable for intelligent hardware platforms. Firstly, the MD-C2F structure is built to capture the subtle movements and posture changes in various aquatic environments, with a light weight achieved by introducing dynamic convolution (DcConv). To make the model perform better in small object detection, the EMA mechanism is incorporated into the MD-C2F. Secondly, the MSI-SPPF module is constructed to improve the performance in identifying the features of different scales and the understanding of complex backgrounds. Finally, the ConCat single-channel fusion is replaced by BiFPN weighted channel fusion to retain more feature information and remove the irrelevant information in drowning features. Relative to the Faster R-CNN, SSD, YOLOv6, YOLOv9, and YOLOv10, the MS-YOLO achieves an average accuracy of 86.4% in detection on a self-built dataset at an ultra-low computational cost of 7.3 GFLOPs.

An expert-novice comparison of lifeguard specific vigilance performance

INTRODUCTION

Lifeguards must maintain alertness and monitor an aquatic space across extended periods. However, lifeguard research has yet to investigate a lifeguard's ability to maintain performance over time and whether this is influenced by years of certified experience or the detection difficulty of a drowning incident. The aim of this study was to examine whether lifeguard experience, drowning duration, bather number, and time on task influences drowning detection performance.

METHOD

A total of 30 participants took part in nine 60-minute lifeguard specific tasks that included 11 drowning events occurring at five-minute intervals. Each task had manipulated conditions that acted as the independent variables, including bather number and drowning duration.

RESULTS

The experienced group detected a greater number of drowning events per task, compared to novice and naïve groups. Findings further highlighted that time, bather number, and drowning duration has a substantial influence on lifeguard specific drowning detection performance.

PRACTICAL APPLICATIONS

It is hoped that the outcome of the study will have applied application in highlighting the critical need for lifeguard organizations to be aware of a lifeguard's capacity to sustain attention, and for researchers to explore methods for minimizing any decrement in vigilance performance.

Virtual reality lifeguarding scenarios as a potential training solution for pool lifeguards

BACKGROUND

Ensuring that pool lifeguards develop the skills necessary to detect drowning victims is challenging given that these situations are relatively rare, unpredictable and are difficult to simulate accurately and safely. Virtual reality potentially provides a safe and ecologically valid approach to training since it offers a near-to-real visual experience, together with the opportunity to practice task-related skills and receive feedback. As a prelude to the development of a training intervention, the aim of this research was to establish the construct validity of virtual reality drowning detection tasks.

METHOD

Using a repeated measures design, a total of 38 qualified lifeguards and 33 non-lifeguards completed 13 min and 23 min simulated drowning detection tasks that were intended to reflect different levels of sustained attention. During the simulated tasks, participants were asked to monitor a virtual pool and identify any drowning targets with accuracy, response latency, and dwell time recorded.

RESULTS

During the simulated scenarios, pool lifeguards detected drowning targets more frequently and spent less time than non-lifeguards fixating on the drowning target prior to the drowning onset. No significant differences in response latency were evident between lifeguards and non-lifeguards nor for first fixations on the drowning target.

CONCLUSION

The results provide support for the construct validity of virtual reality lifeguarding scenarios, thereby providing the basis for their development and introduction as a potential training approach for developing and maintaining performance in lifeguarding and drowning detection.

APPLICATION

This research provides support for the construct validity of virtual reality simulations as a potential training tool, enabling improvements in the fidelity of training solutions to improve pool lifeguard competency in drowning detection.

Active visual search in naturalistic environments reflects individual differences in classic visual search performance

Visual search is a ubiquitous activity in real-world environments. Yet, traditionally, visual search is investigated in tightly controlled paradigms, where head-restricted participants locate a minimalistic target in a cluttered array that is presented on a computer screen. Do traditional visual search tasks predict performance in naturalistic settings, where participants actively explore complex, real-world scenes? Here, we leverage advances in virtual reality technology to test the degree to which classic and naturalistic search are limited by a common factor, set size, and the degree to which individual differences in classic search behavior predict naturalistic search behavior in a large sample of individuals (N = 75). In a naturalistic search task, participants looked for an object within their environment via a combination of head-turns and eye-movements using a head-mounted display. Then, in a classic search task, participants searched for a target within a simple array of colored letters using only eye-movements. In each task, we found that participants' search performance was impacted by increases in set size-the number of items in the visual display. Critically, we observed that participants' efficiency in classic search tasks-the degree to which set size slowed performance-indeed predicted efficiency in real-world scenes. These results demonstrate that classic, computer-based visual search tasks are excellent models of active, real-world search behavior.

Conscientiousness protects visual search performance from the impact of fatigue

Visual search-looking for targets among distractors-underlies many critical professions (e.g., radiology, aviation security) that demand optimal performance. As such, it is important to identify, understand, and ameliorate negative factors such as fatigue-mental and/or physical tiredness that leads to diminished function. One way to reduce the detrimental effects is to minimize fatigue itself (e.g., scheduled breaks, adjusting pre-shift behaviors), but this is not always possible or sufficient. The current study explored whether some individuals are less susceptible to the impact of fatigue than others; specifically, if conscientiousness, the ability to control impulses and plan, moderates fatigue's impact. Participants (N = 374) self-reported their energy (i.e., the inverse of fatigue) and conscientiousness levels and completed a search task. Self-report measures were gathered prior to completing the search task as part of a large set of surveys so that participants could not anticipate any particular research question. Preregistered linear mixed-effect analyses revealed main effects of energy level (lower state energy related to lower accuracy) and conscientiousness (more trait conscientiousness related to higher accuracy), and, critically, a significant interaction between energy level and conscientiousness. A follow-up analysis, that was designed to illustrate the nature of the primary result, divided participants into above- vs. below-median conscientiousness groups and revealed a significant negative relationship between energy level and accuracy for the below median, but not above-median, group. The results raise intriguing operational possibilities for visual search professions, with the most direct implication being the incorporation of conscientiousness measures to personnel selection processes.

Safety optimization in an accident-prone aquatic context: A qualitative study of drowning risk detection by public pool lifeguards

INTRODUCTION

Drownings occur in French public swimming pools (PCP) in spite ofconstant supervision by qualified lifeguards. This studyhypothesizesthat organizational aspects may affect the mission of pool supervision andtakes a systemic approach that views safety as an emergent property, and drowning as a multifactorial and long-term process that is not restricted to the end of an accidental sequence.

METHOD

We conducted a qualitative survey at four municipal poolsin France based on 30 semi-structured interviews.

RESULTS

The findings revealed that several interrelated elements influence the detection by lifeguards of bathers in distress: (a) their training, their conception of the profession, and the reasons that led them to enter this occupation; (b) their representations of the drowning risk and the evaluation of their ability to perceive such situations; (c) the question of regulation; and (d) the methods of dealing with this framework in daily operations, which involves trade-offs and appropriations at each level of the system.

CONCLUSIONS

In order to identify the most significant influences, it seems relevant:(a) to shift the focus away from these professionals; and (b) to prioritize an analysis of the functioning of the risk management system as a whole, not just critical or post-accident periods. In order to reduce the risk to swimmers, it seems preferable to identify the source of the constraints that weigh daily on lifeguards in charge of pool supervision.The functioning of PSP's is the end result of joint regulation processes likely to influence, positively or negatively, the identified feedback loops. An analysis of the pool supervision can help to identify early warning signs of vulnerability (which can sometimes be acted upon at low cost),identify the practical implications, and makepreventive recommendations.

Target-rate effect in continuous visual search

From infrared body temperature surveillance to lifeguarding, real-life visual search is usually continuous and comes with rare targets. Previous research has examined realistic search tasks involving separate slides (such as baggage screening and radiography), but search tasks that require continuous monitoring have generally received less attention. In this study, we investigated whether continuous visual search would display a target-rate effect similar to the low-prevalence effect (LPE) in regular visual search. We designed a continuous detection task for a target feature (e.g., a green color) among items of continuously and gradually changing features (e.g., other colors). In four experiments, we demonstrated target-rate effects in terms of slower hit response times (RTs) and higher miss rates when targets were rare. Similar to regular search, target-rate effects were also observed for relative frequencies across two target features. Taken together, these results suggest a target-rate effect in continuous visual search, and its behavioral characteristics are generally similar to those of the LPE in regular visual search.

Evaluation of strategies to train visual search performance in professional populations

Visual search, the act of finding targets amongst distractors, is central to many professions with life-or-death implications including aviation security, radiology, lifeguarding, military, and more. As such, every effort should be taken to improve visual search performance. One potential path to improvement is to ensure that workforces are optimally trained. Broadly, there are three general components to train: (1) specific use of the machinery and user interface (i.e. 'knobology'), (2) target and distractor identification, and (3) search strategy. The current review considers the cognitive psychology aspects of these three components; each is evaluated in light of short-term and long-term training goals, as well as profession-specific constraints.

Multiple-Target Visual Search Errors

Visual search—the ability to locate visual targets among distractors—is a fundamental part of professional performance for many careers, including radiology, airport security screening, cytology, lifeguarding, and more. Successful execution of visual search in these settings is critically important because the consequences of a missed target can be horrific. Unfortunately, many of these professions place high demands on the people performing the searches, and either the task or the environment (or both) could lead to significant errors. One known source of error that exists across many fields is “multiple-target visual search” errors—a target is less likely to be detected if another target was already found in the same search than if the target was the only one present. These errors have proven to be stubborn and not easily eliminated. This article offers a brief overview of the existing research on multiple-target visual search errors and discusses possible policy implications of the errors for airport security screening. The policy suggestions are based on empirical research, with the hope of providing food for thought on using scientific data and theory to improve performance. Specifically, three policy suggestions are raised: shift screening to a remote location away from the checkpoint, reduce the number of prohibited items to lessen the searchers’ cognitive burden, and emphasize search consistency in the training process. Note that the focus here is on airport security screening, as this is a domain most readers can relate to, but the suggestions can equally apply to many search environments.