This repository contains various scripts and tools for training and evaluating semantic and instance segmentation models. Below is an overview of the repository structure and its contents.
For this project, the following versions were used in the Anaconda environment:
- conda version: 23.5.2
- conda-build version: 3.25.0
- python version: 3.11.3.final.0
Additionally, the following virtual packages were used:
- __archspec=1=x86_64
- __cuda=12.2=0
- __win=0=0
- augmentedDataScripts/: Contains Python scripts for augmenting training data.
- customDataScripts/: Contains scripts for processing training data as desired.
- Mask-RCNN/: Contains training and evaluation code, as well as scripts for running multiple tests sequentially.
- U-Net/: Includes training and evaluation scripts, as well as scripts for running multiple tests sequentially.
- YOLOv8/: Includes training and evaluation code, as well as scripts for running multiple tests sequentially.
- Testdata/: Contains the test datasets.
- Trainingsdata/: Contains the training datasets.
This project is licensed under the GNU General Public License v3.0.
To use the scripts for data augmentation, navigate to the augmentedDataScripts folder and run the desired script with the appropriate parameters. Similarly, processing specific training data can be done in the customDataScripts folder.
Training and evaluation of models can be performed in the respective model folders (Mask-RCNN, U-Net, YOLOv8). Each folder contains detailed instructions on how to run the scripts and set up the environment.
Contributions to this project are welcome. Please refer to the contributing guidelines for more information on how to submit pull requests.
If you encounter any issues or have questions about the repository, please open an issue on GitHub.