Institutional Repository of Vilnius University
Palydovinių duomenų sugretinimas pokyčių aptikimui gamtiniuose arealuose AI pagalba /
Abstract
dc:descriptionDetection of Environmental Changes using Historical Satellite Images with AI Computer vision is tracking more and more attention. Recent advancements already made a huge impact on peoples day to day lives. Self-driving cars, surveillance systems, smart cameras all of these things are not possible without computer vision. Computer vision is currently facing four separate image related problems: image classification, object detection, semantic segmentation, and instance segmentation. In this work, we will be covering instance segmentation which is all previous problems combined. Instance segmentation is a method of AI processing an image to find and individual objects, classifying it and outlining its area at the pixel level. This could be used for various other applications but in this work, we are going to cover its use to find deforestation areas in satellite images and calculating the area of the deforestation region. One of the main purposes of this work is to create a software tool to detect environmental changes in forest areas. For this purpose, MRCNN is going to be used which does instance segmentation by processing images. After images were processed by AI the output is going to be projected onto the map and its area will be measured. By testing trained the AI following results were noticed: it detects deforestation with the precision of 83%, the loss of the calculated areas is 7% and precision to detect environmental changes in a given time is 84%.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2020
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Česnulevičius, Rokas,
- Contributors dc:contributor
-
- Čivilis, Alminas
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD81918289&prefLang=en_US