Dataset Details
- Training Data: 78,200 images (1,700 per character)
- Test Data: 13,800 images (300 per character)
- Total Images: 92,000
- Image Specs: Each image is 32×32 pixels, with 3 color channels.
Context
The challenge lies in differentiating similar-looking characters, especially given the variability of handwritten samples. The dataset aims to address the complexities in recognizing Devanagari script, with a focus on overcoming issues related to sloppy or illegible writing.