This repository is for CoinNet: Deep Ancient Roman Republican Coin Classification via Feature Fusion and Attention (CoinNet) introduced in the following paper

Hafeez Anwar, Saeed Anwar, Sebastian Zambanini, Fatih Porikli, "CoinNet: Deep Ancient Roman Republican Coin Classification via Feature Fusion and Attention", Pattern Recognition, 2021, PDF and arXiv

The model is built in

• PyTorch 1.1.0,
• Ubuntu 14.04/16.04 environment
• Python 3.7.5
• CUDA9.0
• cuDNN5.1

1. Introduction
2. Network
3. Test
4. Dataset
5. Results
6. Citation

We perform classification of ancient Roman Republican coins via recognizing their reverse motifs where various objects, faces, scenes, animals, and buildings are minted along with legends. Most of these coins are eroded due to their age and varying degrees of preservation, thereby affecting their informative attributes for visual recognition. Changes in the positions of principal symbols on the reverse motifs also cause huge variations among the coin types. Lastly, in-plane orientations, uneven illumination, and a moderate background clutter further make the task of classification non-trivial and challenging. To this end, we present a novel network model, CoinNet, that employs compact bilinear pooling, residual groups, and feature attention layers. Furthermore, we gathered the largest and most diverse image dataset of the Roman Republican coins that contains more than 18,000 images belonging to 228 different reverse motifs. On this dataset, our model achieves a classification accuracy of more than 98% and outperforms the conventional bag-of-visual-words based approaches and more recent state-of-the-art deep learning methods. We also provide a detailed ablation study of our network and its generalization capability.

The following figure shows the architecture of our network

Our model highlighting the Compact Bilinear Pooling, residual blocks, skip connections, and feature attention. The green and yellow cubes indicate the embedded features via CNN networks.

1. Download the trained models for our paper and place them in '..'. The model can be downloaded from Google Drive. The size is 409MB

2. Cd to 'CoinNet_Model', run the following scripts. You can use the following script to test the algorithm

   #CoinNet
   CUDA_VISIBLE_DEVICES=0 python test.py

The dataset can be downloaded from Google Drive. The size of the dataset is 8.65GB

If you find the code helpful in your resarch or work, please cite the following papers.

@article{Anwar2021CoinNet,
    title = {Deep Ancient Roman Republican Coin Classification via Feature Fusion and Attention},
    journal = {Pattern Recognition},
    pages = {107871},
    year = {2021},
    issn = {0031-3203},
    doi = {https://doi.org/10.1016/j.patcog.2021.107871},
    url = {https://www.sciencedirect.com/science/article/pii/S0031320321000583},
    author = {Hafeez Anwar and Saeed Anwar and Sebastian Zambanini and Fatih Porikli},
}