This repository contains a dataset and an experiment notebook for a convolutional classification model trained to recognize six basic troop types from Mount & Blade II: Bannerlord. The project was completed as part of the "Neural Networks and Deep Learning" course at the Faculty of Informatics Pula. The dataset was collected and labeled by the project author.
Bannerlord.ipynb: Jupyter notebook with the experimental code for data loading, training, validation, and basic evaluation/visualizations.Dataset/: image dataset split intotraining/,validation/, andtest/folders. Each split contains class subfolders.Documentation.pdf: additional notes and documentation produced for the project.
The repo follows a standard image-classification layout:
Dataset/
training/
Aserai Recruit/
Battanian Volunteer/
Imperial Recruit/
Khuzait Nomad/
Sturgian Recruit/
Vlandian Recruit/
validation/
(same class subfolders)
test/
(same class subfolders)
Each class folder holds images for that troop type. This structure is compatible with common PyTorch ImageFolder and TensorFlow image_dataset_from_directory utilities.
- Clone the repository
git clone https://github.com/Sebastijan-Dominis/bannerlord_classification_model
cd bannerlord_classification_model- Create and activate a virtual environment:
If using Anaconda:
conda create -n "bannerlord_classification_model" python=3.11
conda activate bannerlord_classification_model- Install the project's dependencies from
requirements.txt:
pip install -r requirements.txt- Use the provided Jupyter notebook:
The notebook contains the code to load the images from
Dataset/, define a model, train, and evaluate.
- The dataset was collected by the repository author for educational purposes. Redistribution and reuse are subject to the terms of the MIT License.
- See the
LICENSEfile for the full text.
- Author: Sebastijan Dominis
- Contact: sebastijan.dominis99@gmail.com