Add Benchmark Table for OCR Models in README

[ishika24755]

Description / Problem:
The README currently states:

“All models, regardless of the approach used, achieve over 90% accuracy.”

However, it does not provide a structured comparison of the different models (CRNN, Vision Transformer, Self-Supervised Learning). This makes it harder for users to understand the differences between the approaches or compare their performance.

Proposed Improvement:
Add a benchmark table in the README showing each model’s performance, including:

Model | Dataset | Accuracy | Notes -- | -- | -- | -- CRNN | letters | 91% | Example note Vision Transformer | letters | 93% | Example note Self-Supervised Learning (SSL) | letters | 92% | Example note
You can also optionally include sample outputs or screenshots if available.

Why This Matters:

Provides clear and structured comparison of models

Improves readability and clarity for users

Adds professional and academic credibility to the repository

[Deepthanush1337]

This would be really useful. Right now the README just says everything is above 90% which doesn't tell you much about how the different approaches compare.

I can put together a benchmark table if you want. I was going to go through each model's training scripts and notebooks to pull out the actual reported metrics rather than just guessing numbers. Would something like model name, dataset, CER, WER, and any notable details work?