Spot! Online Help

The spot pattern recognition algorithms used in whale shark mark-recapture (Modified Groth and I3S) assume a flat, two-dimensional surface when analyzing the relationships between spots. While each algorithm has some tolerance for skew in an image, both quickly degrade in their ability to match identical patterns as the angles between those patterns increases.

Spot! allows you to map a skewed 2D image to a 3D whale shark model and obtain a properly-oriented left- or right-side pattern for use with the Modified Groth and I3S algorithms.This perspective correction has been used in the ECOCEAN Library to match images taken from very extreme angles to previously tagged whale sharks. For example, Spot! was used to make this match for shark M-025 in the ECOCEAN Library.

 

Spot! Requirements

Spot! requires Java 5 or Java 6 on your computer. The Java Runtime Environment (JRE) can be downloaded from Sun Microsystems at:

http://www.java.com

Loading Spot!

Spot! can be loaded from within the Client Software page of the ECOCEAN Library or directly from this link:

http://www.whaleshark.org/spot/spot.jnlp

Basic Spot! Instructions

There are just a few basic steps to using Spot!.

  1. Load the skewed image into Spot! using the Open button.
  2. Scale the image to the 3D model by holding down the middle mouse button and moving the mouse forward and backward.
  3. Move the loaded image left and right by holding down the right mouse button.
  4. Manipulate the 3D model by holding down the left mouse button on its edges and moving the mouse up, down, left, and right.
  5. Repeat steps 2-4 to align the guidelines on the 3D to these features on the 2D image: 5th gill, lowest lateral line, and vertebral column.
  6. Once aligned, click the Map button to map the 2D image to the 3D model.
  7. Click the Export button to save the properly aligned image for spot mapping with ECOCEAN Interconnect.

The Quickstart video demonstrates these steps.

Spot! FAQ

1. How accurate is Spot?

We're still testing it. So far, we've made a lot of very strong matches in our test cases using real-world data. Spot! has exceeded our expectations in its ability to match patterns from skewed images to properly-oriented patterns. As with any software tool, the result is as good as the user, and Spot! requires patient, careful alignment between 3D model and 2D image. At this time, we're using Spot! to go back through encounter reports and identify sharks from images that were previously unsuitable for existing pattern recognition algorithms. Statistical tests of Spot!'s accuracy will be performed in the near future. But again...it's always going to be only as good as its user.

2. Can I use Spot! to identify new whale sharks from skewed images?

That's a longer discussion about misidentification in mark-recapture population modeling. At this point, Spot! is NOT being used to identify new whale sharks in the ECOCEAN Library. However, we may choose to do so in the future. We currently use it only to identify previously marked whale sharks. In general, we HIGHLY recommend the tandem use of both the Modified Groth and I3S pattern recognition algorithms in addition to rigorous peer review to reduce misidentification.

3. Can Spot! be used for other species?

Yes! Spot! can load other 3D models (.obj file format) and map images to them. This allows researchers for other spotted species to use image catalogs and pattern recognition algorithms where the observer cannot orient the camera correctly to the animal, such as when camera traps are used. If you are interested in using Spot! and the dual pattern recognition algorithms of the ECOCEAN Library, please contact webmaster at whaleshark dot org.

4. Is Spot! open source?

It will be as soon as our initial tests are complete. For now, it is freeware under license. Use it as you like.

5. Is a single whale shark model representative of the long-term morphology of an individual?

Probably not for very small (rare) and very large whale sharks, such as those found in the Galapagos Islands. The truncated 3D whale shark model used in Spot! is based on the proportions of average sized whale sharks, generally about 7 meters in total length (TL). We assume changes are proportional in all three dimensions for smaller and larger sharks (generally 5 m to 10 m in TL) within the range of most ecotourism activity. As our understanding of whale shark biology increases, we may use multiple models for different stages of growth.

6. Why is this model truncated at the front and back? Where are the fins?

The model focuses only on the fiducial region used for spot pattern recognition of whale sharks: left- and right-side patterning behind the gills. This area does not distort as much as many other regions of the shark during normal motion. By using a truncated model, we focus the eye on only the region of interest and consider only he region of interest.