Scenario A, phase 6

Robin, your trusted colleague in IT, has combined the video recordings of the six drones into a single massive patchwork TIFF image (called an orthomosaic) of the entire downtown area. Each pixel can be roughly mapped into a latitude-longitude pair.

Hands-on option

Basic stage for the hands-on option

Formulate a plan that indicates what to do with the orthomosaic image and how to advance your plan of detecting and repairing potholes and other pavement defects. You can include some open-licence photos from the internet as examples illustrating the proposed steps and the expected outcomes.

In-depth stage for the hands-on option

(This task is recommended for those who have already taken CCCS 610.) Carry out some image processing in Python to highlight a few elements of interest in an orthomosaic (or aerial) image of your choice. There are not many options online with free usage rights, but you could, for example, highlight the walls of the structure in this Wikimedia image of the Lancia ruins.

Conceptual option

Basic stage for the conceptual option

Browse through Computer Vision: Principles, Algorithms, Applications, Learning (a comprehensive computer-vision textbook by E. R. Davies and E. Roy, 2018), and draw a flow chart of what you understand a typical process of computational intelligence to be when the input data consists of one or more images, and the output is some sort of recommendation regarding the content of the input data.

In-depth stage for the conceptual option

Investigate how multiple images can be combined into a large orthomosaic image (for example, this one on the Geonadir website). Again, draw a flow chart to illustrate the steps of the process. At each step, attach a note indicating how computationally challenging it appears to be (both how hard it is to design an algorithm for that step and how slow the algorithm would be in actually completing the step).