This is the code for my Bachelors thesis at the IT University of Copenhagen.

In this project I explore the fine-tuning of object detectors in deep sea environments, and how they can be adapted for deployement in underwater vehicles.

Personally, I use mamba, but conda will work as well. To create a new env with all the dependencies use the following command:

mamba env create -f env.yaml

To update the env is new packages are added to yaml file:

mamba env update -n ocean --file env.yaml

• make add_datasets Add datasets to FiftyOne.
• make launch_51 Launch FiftyOne app. You will be prompted to choose which dataset you want to look at, AUV, ROV or both merged into one.
• The app will launch in the browser automatically. It can be accessed at http://localhost:5151/
• The process shuts down when you exit the app.

Useful commands:

• fiftyone datasets list list existent datasets.
• fiftyone datasets delete <name1> <name2> delete datasets.