This project uses classical computer vision techniques to generate objects for human labeling from aerial drone footage. Target labels are dolphins. This repo also contains a GUI used to label the generated objects.
python dolphins.py [-h] [-f FILE] [-fo FOLDER] [-d{d}] [-np] [-n NCORES]
- -h, shows the help screen.
- -f, Path to single image to be analysed.
- -fo, Path to folder of images to be analysed.
- -d, Display debug info. Has differnet levels of debug information, e.g -dd shows more debug info than -d.
- -np, Supress default plot output.
- -n, Specify the number of cores to use. Default is 1.
If you are using anaconda to manage your python enviroment then use the following to install all dependancies
- conda env create -f enviroment.yml
- opencv
- numpy
- scikit-image=0.16.2
- matplotlib
- python=3.7
- tensorflow