Hand-written-characters-recognition-via-deep-metric-learning

Code for the paper 'Hand Written Characters Recognition via Deep Metric Learning'
Find the paper at: CLICK HERE
The dataset WILL BE AVAILABLE SOON from: CLICK HERE.

Dataset Description

Computer vision community has extensively used MNIST dataset in different applications including similarity. However, the dataset has only few seemingly similar classes, making it less effective for deep metric learning methods. In this paper, a new handwritten dataset named Urdu-Characters is created in a similar way as MNIST dataset.

Dataset Images

Contact

Please contact for any questions/queries: snawaz@uninsubria.it

Citation

If you use the dataset and/or the code, consider citing the paper:

@INPROCEEDINGS{2018:nawaz:DAS,
  author    ={S. Nawaz, A. Calefati, N. Ahmed, I. Gallo},
  booktitle ={13th IAPR International Workshop on Document Analysis Systems (DAS)},
  title     ={Hand Written Characters Recognition via Deep Metric Learning},
  year      ={2018},
  volume    ={05},
  pages     ={417-422},
  keywords  ={Handwritten dataset; Deep metric learning; Triplet network; Siamese network;},
  doi       ={10.1109/DAS.2018.18},
  ISBN13    ={978-1-5386-3346-5/18},
  month     ={Apr},
}