Pompey21 / Don-t-Let-the-Bark-Beetle-Bite-You

For my dissertation I worked on developing a Computer Vision algorithmic solution for detecting bark beetle infestations in forests through analysing satellite images.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Don-t-Let-the-Bark-Beetle-Bite-You

The supporting material for the dissertation includes the code (1), majority of the data used (2), newly developed data-sets (3), and results (4) of the project. All of these are accessed through their own folders.

Here is the list of all the folders and what they contain:

Code

There are five .ipynb files available in this folder.

  • model.ipynb : contains the whole model
  • milanov_vrh_test.ipynb : tests conducted on the region of Milanov Vrh
  • parameter_tunning.ipynb : operations for the search of best parameter combination for cloud-masking component
  • infected_regions_test.ipynb : tests conducted on the regions that had been attacked by the bark beetle
  • cloud_masking_evaluation.ipynb : tests on the performance of the cloud masking algorithm

Data

Contains a set of data for regions in Slovenia and the ones in Croatia. These can be accessed through their respective folders.

  • Slovenia : Contains a document where all the regions are stated in coordinates - the reports have been preprocessed and the coordinates have been taken out. As the reports were in Slovenian language they were not submitted here.
  • Cabar : Contains two files, one with the coordinates for the two damaged forest areas in the Cabar region in Croatia, and the second one with the raw data for the Milanov Vrh area (this one is in Croatian!)

New Data Sets

Contains three files:

  • Locations of Regions contains only the coordinate locations of the regions that had been used in this study - the ones that have been attacked.
  • Locations of Regions and Years contains the same information as the previous file however it also provides the year prior and the year of the attack so that the images can be accessed more easily.
  • Milanov Vrh contains raw data for the region of Milanov Vrh.

Results

This folder contains only the test results for the regional tests - if the model can recognise the bark beetle attack - and the results of different combinations of parameters for the cloud-masking algorithm.

Notes

Please note that the images used have not been put as part of the supporting material as this required much more organisation. If you do want to run particular tests, please contact me on marko.mekjavic@gmail.com and I will send you the necessary instructions and missing materials.

About

For my dissertation I worked on developing a Computer Vision algorithmic solution for detecting bark beetle infestations in forests through analysing satellite images.


Languages

Language:Jupyter Notebook 97.5%Language:TeX 2.4%Language:Python 0.1%