Overview

This repository is for the benthic species identification task for the ROV MATE Competition of 2019. Created by the DVC ROV team of 2019.

Setup

# Clone the repo and change directories into the folder
git clone https://github.com/dvc-rov/shape-detection.git
cd shape-detection

# Create the environment from the .yml file and activate it
conda env create -f environment.yml
conda activate dvc-rov

Usage

1. src/detect_shapes.py - Contains the entire logic for detecting shapes (i.e. benthic species)
2. src/test_pictures.py - Test against the two example pictures in the test/ folder. You can specify an image file for recognizing, or omit it to default to loading sample_complex.png. Press 'q' at any time to stop the script.
3. src/test_webcam.py - Test against your webcam. You can specify a video port number or omit it to load the first one it can find. Press 'q' at any time to freeze the camera frame and again to stop the script.
4. src/test_screencap.py - Test against your screen. It'll only look at the left half of your screen to avoid the recursive screen effect. Press 'q' at any time to freeze the screen frame and again to stop the script.
5. src/test_qgc.py (NOT TESTED) - Test against the QGC camera feed. Press 'q' at any time to freeze the screen frame and again to stop the script.

NOTE: DO NOT click the 'X' on the window, since it won't terminate the python script and you won't be able to Ctrl+C the script. Press 'q' to gracefully end the script (twice if testing with the webcam or screen).