KM-1/BERBER-MNIST-dataset

The BERBER-MNIST Dataset is a set of handwritten uppercase character derived from the EMNIST Dataset and compatible with the MNIST dataset. The compatibility of this dataset with the MNIST paradigm makes its easy to use without changing your algorithms. The BERBER-MNIST Dataset ======================== What is it? ----------- The BERBER-MNIST Dataset is a set of handwritten uppercase character of Amazigh script. Derived from the EMNIST Dataset and **compatible with the All MNIST algorithms**. Formats ------- The dataset is provided in .csv format. You can convert .csv file into images .png format using the following python notebook script: *csv2png.ipynb*. Dataset Summary --------------- There are two different version of Amazigh Script provided in this dataset. A short summary of the dataset is provided below: | Latin-version | Tifinagh-version | | --- | --- | | 247,985 characters | 25,740 characters | | 33 unbalanced classes. | 33 unbalanced classes | | A,Ԑ,B,C,Č,D,Ḍ,E,F,G,Ǧ,H,Ḥ,I,J,K,L,M,N,Q,Γ,R,Ṛ,S,Ṣ,T,Ṭ,TT,U,W,X,Y,Z,Ẓ | ⴰ,ⴱ,ⵛ,ⴷ,ⴹ,ⵄ,ⴼ,ⴳ,ⵖ,ⴳⵯ,ⵀ,ⵃ,ⵊ,ⴽ,ⴽⵯ,ⵍ,ⵎ,ⵏ,ⵇ,ⵔ,ⵕ,ⵙ,ⵚ,ⵜ,ⵟ,ⵡ,ⵅ,ⵢ,ⵣ,ⵥ,ⴻ,ⵉ,ⵓ | | | | Author ------- K.Muqran at CRLCA Files ----- The folder consists of the following files: ```bash BERBER-MNIST │ Latin-version.csv | Tifinagh-version.csv | csv2png.ipynb │ README.md ``` ``` kaggle.com doi.org ```