The goal is to detect invasive plant species in google street view images using a trained convolutional neural network (CNN).
The gsvannotation folder contains code for a web based annotation tool. Initially, the tool allows you to enter a lat/lon , if there is a GSV panorama for this lat/lon, it is downloaded and you can draw bounding boxes around invasive plant species. The goal is to use this tool to collect training data and/or test trained models and verify their performance.
To build and run the project you will need the following
pip install numpy
pip install pillow
To build the annotation tool, cd into the gsvannotation folder and run mvn install
. Maven should download all the required dependencies. Dependencies include:
Alternatively, you can use an IDE such as Eclipse ( www.eclipse.org ). In eclipse, import the project as a maven project.
To run the project
mvn install
java -jar target/gsvannotation-jar-with-dependencies.jar [dbhost] [dbusername] [dbpassword] [dbport]
. Assuming the database is on the same machine, it would be something like java -jar target/gsvannotation-jar-with-dependencies.jar localhost someuser supersecretpassword 3306
In eclipse, you can also run/debug the Main.java class as a plain java program. You will have to edit the run/debug configuration to include the command line arguments. Eclipse makes it easy to debug changes, once you run the program, any changes you make are usually available instantly. In some cases it may be necessary to restart the program.