Batman

Trajectory shape matching (Batman)

$Description: C:\Users\radum\AppData\Local\Microsoft\Windows\INetCache\Content.Word\matching.jpg$

Input:

One GPS trajectory and a set of shapes (silhouettes).

Output:

The best match + accuracy + is the matching significant?

Why?

- Think gamification = one more way to attract users to do sports and record trajectories!

- Another reason is that popular sports tracking applications require users to enter a description for their trajectories. Automatic descriptions are sometimes generated: “Afternoon ride”, “Morning run”. These can be made more exciting if the trajectory happens to match a certain shape.

- Unlike the other tasks, this one is research-oriented and scientific paper can come out of it. It is essentially a generalization of:

R. Mariescu-Istodor, P. Heng and P. Fränti, "Roundness measure for GPS routes"

Conference on Location-based Services (LBS), Zürich, Switzerland, January 2018

How?

This computation should be done on the server. Only the result will be sent to the web / mobile users. You may use any of the following programming languages:

Java, PHP, C, C++, Python

There are many ways of solving this task. One way is to first preprocess the shapes by extracting the contours using OpenCV. Then, shape similarity may be computed between the trajectory and all contours. Consider how the method can be scale and rotation invariant. Defining a different similarity measure may be needed entirely.