Deep learning project to identify characters of the urdu alphabet.
Created a custom dataset consisting of 1550 images of all 39 alphabets of the urdu language (~40 images per class).
Utilised pytorch to create a VGG-16 and AlexNET models and compared the results between the 2.
Results:
| Model | Accuracy | F1 Score | Memory (MB) |
|---|---|---|---|
| VGG - 16 | 96.15% | 0.964 | 97.86 |
| AlexNet | 90.71% | 0.906 | 26.01 |