README.md

metadata

library_name: transformers
tags:
  - Maths
  - Handwritten
  - OCR
  - Transformers

Model Description

This model performs OCR on hand-written calculus formulas, extracting text with high accuracy

Usage

Installation

pip install unsloth transformers torch pillow

Inference Code

from unsloth import FastVisionModel
import torch
from transformers import AutoModel
import os

os.environ["UNSLOTH_WARN_UNINITIALIZED"] = '0'

# Load the model
model, tokenizer = FastVisionModel.from_pretrained(
    "karankulshrestha/formula_ocr",
    load_in_4bit = False,
    auto_model = AutoModel,
    trust_remote_code = True,
    unsloth_force_compile = True,
    use_gradient_checkpointing = "unsloth",
)

# Enable fast inference
FastVisionModel.for_inference(model)

# Run OCR on an image
result = model.infer(
    tokenizer,
    prompt="\nFree OCR.",
    image_file="path/to/your/image.jpg",
    output_path="./output",
    base_size=1024,
    image_size=640,
    crop_mode=True,
    save_results=True,
)

print("OCR Result:", result)

Output

Citation

@misc{formula_ocr},
  author = {Karan Kulshrestha},
  title = {Hand-written OCR Model},
  year = {2025},
  publisher = {HuggingFace},
  url = {https://huggingface.co/karankulshrestha/formula_ocr}
}