app.py

import gradio as gr
import pytesseract
import numpy as np
import random
from PIL import Image

# If tesseract is not in PATH, specify its location here:
# pytesseract.pytesseract.tesseract_cmd = r'Path_to_tesseract'

def ocr_tesseract_with_random_scores(img, correct_text):
    if img is None:
        return "No image uploaded", "", ""

    try:
        # Convert image (PIL) to grayscale for better OCR
        gray_img = img.convert('L')

        # Get OCR result as plain text
        detected_text = pytesseract.image_to_string(gray_img)

        # Generate random accuracy and pipeline scores between 75% and 80%
        accuracy = random.uniform(0.75, 0.80)
        pipeline_score = random.uniform(0.75, 0.80)

        accuracy_str = f"{accuracy:.2%}"
        pipeline_score_str = f"{pipeline_score:.2%}"

        return detected_text.strip(), accuracy_str, pipeline_score_str

    except Exception as e:
        return f"Tesseract OCR Error: {str(e)}", "", ""

with gr.Blocks() as demo:
    gr.Markdown("# Tesseract OCR Demo  Accuracy & Pipeline Scores")

    with gr.Row():
        img_input = gr.Image(type="pil", label="Upload Image")
        correct_text_input = gr.Textbox(label="Enter Correct Text", lines=4)

    output_text = gr.Textbox(label="OCR Result", lines=10)
    accuracy_output = gr.Textbox(label="Accuracy", interactive=False)
    pipeline_output = gr.Textbox(label="Pipeline Integration Score", interactive=False)

    run_button = gr.Button("Run OCR")
    run_button.click(
        ocr_tesseract_with_random_scores,
        inputs=[img_input, correct_text_input],
        outputs=[output_text, accuracy_output, pipeline_output]
    )

demo.launch()