Visual search performance in 'CCTV' and mobile phone-like video footage

Area of Science:

• Cognitive Psychology
• Human-Computer Interaction
• Forensic Science

Background:

• Identifying unfamiliar individuals in video footage is crucial for law enforcement and security.
• Current visual search accuracy for unfamiliar targets in video is often low.
• The effectiveness of using multiple target images versus a single image is not well-established.

Purpose of the Study:

• To investigate whether using four images of a target person improves accuracy in visual search tasks compared to using one image.
• To examine this effect in simulated CCTV footage and body-camera/mobile phone recordings.
• To identify potential biases in responding based on the number of target images provided.

Main Methods:

• Two real-world visual-search experiments were conducted.
• Participants searched video footage for an unfamiliar target person.
• The number of target images (one vs. four) and video recording perspective (overhead CCTV vs. shoulder-height body-camera) were manipulated.

Main Results:

• Using four target images generally led to higher accuracy in visual search tasks.
• This improvement in accuracy reached statistical significance in the shoulder-height video condition (Experiment 2).
• A conservative response bias was observed, with participants more likely to report the target absent when given only one image.

Conclusions:

• Providing multiple images of a target person offers an advantage for improving accuracy in video visual-search tasks.
• The benefit of multiple images is more pronounced in realistic, shoulder-height recording scenarios.
• These findings have implications for optimizing search protocols in security and law enforcement.