Assisting Search and Rescue through Visual Attention

You're lost in the wilderness and need to be found. A search and rescue team send an Unmanned Aerial Vehicle (UAV) ahead to find you. How can we modify the aerial photography captured by this UAV to help the team find you as quickly and accurately as possible?

Datasets from the Research

This webpage is intended as an appendix to the thesis entitled “Assisting Search and Rescue through Visual Attention”, available below.

These appendices include all the data analysed from the four experimental investigations:

  1. Serial Visual Presentation (chapter 4, identified by svp, codename IIMC, trials 1–6)
  2. Segmentation-based Inspection (chapter 5, identified by seg_singular and seg_clustered, code names RSVPMSA and RSVPMSB)
  3. Enlargement-based Inspection (chapter 6, identified by enl, codename RSVPE)
  4. Gaze-contingent Displays (chapter 7, identified by gcd, codename IIMC, trials 7 and 8)

Each archive available from this page is in an open format and includes a document describing the data enclosed. If you require any help processing this data, or wish to collaborate, please feel free to email me at james@keot.co.uk.

Download “Assisting Search and Rescue through Visual Attention” (.pdf)

Supporting Materials

The forms and protocols used for the experimental investigations are included here. These materials were used as part of the ethics approval applications for these experiments.

Each archive includes a README.md file to explain the PDFs included.

Download Exemplar Filled Forms (.tar.gz 1.3 MB)

Download Exemplar Blank Forms (.tar.gz 256 kB)

Download Exemplar Protocols (.tar.gz 208 kB)

Naturalistic Stimuli

The aerial photography (a.k.a. terrain strips) used within the experimental investigations was converted from the Natural Resource Information System of the Montana Geographic Information Clearinghouse.

The “Terrain Strip Archive” includes the six original terrain strips that were used in each experimental investigation. These are identified by the letters A–F, with terrains A and B consisting of scrubland and terrains E and F consisting of dense forest. Terrain maps C and D are somewhere between the two extremes of scrubland and dense forest. Two versions of the each terrain strip are available; one with clustered targets, and one with singular targets.

GitHub repositories (source code) are available for both MapTiler which displays the stimuli and MapCutter that prepares the stimuli from the source terrain strips.

Download Terrain Strip Archive (.tar.gz 199 MB)

View MapTiler GitHub Repository (GitHub)

View MapCutter GitHub Repository (GitHub)

Download Target Locations (.tar.gz 4.3 kB)

Eye-movement Records

The eye-movement behaviour of each subject was recorded using a passive LC Technologies eye-gaze tracker system, which records the position of the eye relative to the display screen coordinates every 17 ms (60 Hz). The records available from this appendix include the default columns for each sample as well as some extra columns for relating this dataset to the experimental investigations described in the thesis and documentation.

Download Eye-movement Records (.tar.gz 77 MB)

Associated Publications

A Comparison of Image Inspection Modes for a Visual Search and Rescue Task

A Comparison of Image Inspection Modes for a Visual Search and Rescue Task

Journal of Behaviour & Information Technology, Taylor & Francis (2013)

Download Article (.pdf 1.6 MB) Download Citation (.bib 288 B)

Gaze Behaviour during a Challenging Visual Search Task

Gaze Behaviour during a Challenging Visual Search Task

An ICA Commission on Cognitive Visualisation, Workshop on Eye Tracking: Why, When and How? (2013)

Download Article (.pdf 461 kB) Download Slides (.pdf 2.6 MB) Download Citation (.bib 317 B)

An Interface for Visual Inspection based on Image Segmentation

An Interface for Visual Inspection based on Image Segmentation

ACM Conference on Advance Visual Interfaces (AVI 2012)

Download Article (.pdf 2.1 MB) Download Poster (.pdf 406 kB) Download Citation (.bib 349 B) Watch Preview Presentation (YouTube)

Gaze-Contingent Enhancements for a Visual Search and Rescue Task

Gaze-Contingent Enhancements for a Visual Search and Rescue Task

ACM SIGGRAPH Conference on Applied Perception in Graphics and Visualization (APGV 2011)

Download Article (.pdf 186 kB) Download Poster (.pdf 636 kB) Download Citation (.bib 363 B)

Detecting Search and Rescue Targets in Moving Aerial Images Using Eye-Gaze

Detecting Search and Rescue Targets in Moving Aerial Images Using Eye-Gaze

Conference on Communication by Gaze Interaction (COGAIN 2009)

Download Article (.pdf 1 MB) Download Citation (.bib 282 B)

Subject Training

Segmentation-based Inspection: Introduction Video

Segmentation-based Inspection: Introduction Video

Watch the Introduction for Segmentation-based Inspection (YouTube)

Enlargement-based Inspection: Introduction Video ‘A’

Enlargement-based Inspection: Introduction Video ‘A’

Watch Introduction ‘A’ for Enlargement-based Inspection (YouTube)

Enlargement-based Inspection: Introduction Video ‘B’

Enlargement-based Inspection: Introduction Video ‘B’

Watch Introduction ‘B’ for Enlargement-based Inspection (YouTube)