This repo consists of work on Sea Floor Sampling using machine learning and computer vision techniques, in which there are two main objectives:
There are a couple ways you can do this.
git clone https://github.com/ioos/seafloor-sampling-ml.git
anaconda
environment to run the code.new environment
on your system. To create a new environement, check the steps below
conda create --name <name of the envorinment you want to create> python=<python version you want to install>
conda create --name SeaFloorSampling python=3.8
conda activate <name of the environment you created
conda activate SeaFloorSampling
After we have installed anaconda, and inside the environment we created, we need to setup libraries to use/run the code.
README.md
file. cmd
. This will open the terminal.conda activate <name of the environment you created
. For example, conda activate SeaFloorSampling
.pip install -r requirements.txt
to install all the necessary libraries to run the code further on.This program measures how blurred a given image is. It uses Fourier Transform to measure the level of blur. If the image is too blurred, we can't perform any kind of analysis on it.
blur_detection
folder and have the images in place, run the following code.
python3 blur_level_analyzer.py -f <folder destination which contain other folders (eg. /images)> -c "<name of the final csv file you want to have>"
. For example, python3 blur_level_analyzer.py -f images -c SeaFloorSampling
This program measures what the area of the image being analyses is. This will be crucial when the report is made finally.
laser_area_computation
folder and have the images in place, run the following code.
python3 laser_detection_area_computation.py -f <folder destination which contain other folders (eg. /images)> -c "<name of the final csv file you want to have>"
. For example, python3 blur_level_analyzer.py -f images -c AreaComputation_SeaFloorSampling
This program analyses the images and picks out images with 100% sand or mud images.
Coming soon...
MIT © [IOOS]()