Context This Dataset is just the image representation of this [Dataset][1]. **I had no role in constructing the original images and all the credit goes to the original authors of the dataset Ahmed El-Sawy, Mohamed Loey, Hazem EL-Bakry, and their [paper][2] "Arabic Handwritten Characters Recognition using Convolutional Neural Network, WSEAS, 2017" .** The original dataset has the images as an array of size (32,32,1) this dataset only converts the arrays to their respective jpg images which I hope would make it easier for others to create more notebooks using the dataset. Also in the starter kernel for this dataset, **I obtained _as far as i know_ a new SOTA accuracy score for the test set ~98.6% improving upon the original result from the original paper ~94.9%.** The notebook is using a pre-trained [resnet50 from Pytorch models][3] and using the [Fast.ai][4] library which makes the training and validation super easy and fast (< 10 minutes on the GPU kernel). Content The dataset is composed of 16,800 characters written by 60 participants, the age range is between 19 to 40 years, and 90% of participants are right-hand. Each participant wrote each character (from ’alef’ to ’yeh’) ten times on two forms. The database is partitioned into two sets: a training set (13,440 characters to 480 images per class) and a test set (3,360 characters to 120 images per class). Writers of the training set and test set are exclusive. The ordering of including writers to test set are randomized to make sure that writers of the test set are not from a single institution (to ensure variability of the test set). Acknowledgements All credit goes for the original authors of this dataset who made available a great dataset that is essential for anyone looking into Arabic character recognition and I hope to see more like it in other fields of the Arabic literature. Inspiration I'm looking forward to seeing who the community can explore this dataset more and improve the accuracy scores. [1]: https://www.kaggle.com/mloey1/ahcd1/home [2]: http://www.wseas.org/multimedia/journals/computerresearch/2017/a045818-075.pdf [3]: https://pytorch.org/docs/stable/torchvision/models.html#id3 [4]: https://docs.fast.ai/