This repository contains all the necessary code and documentation of the algorithm for the light trap and computer vision system to detect and classify live moths. It contains instructions and code for both training and data-processing.
Paper: https://www.biorxiv.org/content/10.1101/2020.03.18.996447v1
The following dependencies must be installed.
| Dependency | Version |
|---|---|
| scikit_image | tbd |
| numpy | tbd |
| scipy | tbd |
| Pympler | tbd |
| tensorflow | tbd |
| Pillow | tbd |
| PyQt5 | tbd |
| OpenCV | tbd |
| Seaborn | tbd |
| Scikit learn | tbd |
- Install the dependencies and create the environment using the provided OS specific environment file with the command "conda create --name myEnv --file ENV_FILE.txt
- Activate the enviorement using the command "activate myEnv"
Start the programs by running the files MCC_gui.py or MCC_algorithm.py in the code directory with the command "python MCC_gui.py" or "python MCC_algorithm.py".
New models can be trained using the provided hp_param_training jupyter notebook script.
Edit the script before training, providing a data path, logging path, model save path and edit the steps per epoch.
Tensorboard command: tensorboard --logdir "LOG_PATH_HERE" --reload_multifile=true
Comming soon.
Jakob Bonde Nielsen or Kim Bjerge