Update app.py

app.py

@@ -8,6 +8,18 @@ import gradio as gr
 import requests
 import json
 from dotenv import load_dotenv
+import spaces
+from transformers import AutoModel, AutoTokenizer
+from PIL import Image
+import numpy as np
+import os
+import base64
+import io
+import uuid
+import tempfile
+import time
+import shutil
+from pathlib import Path
 
 # Attempt to install pytesseract if not found
 try:
@@ -134,55 +146,91 @@ def dummy_analyze(ingredients_list, health_conditions=None):
     return report
 
 # Function to extract text from images using OCR
-def extract_text_from_image(image):
+tokenizer = AutoTokenizer.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True)
+model = AutoModel.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True, low_cpu_mem_usage=True, device_map='cuda', use_safetensors=True)
+model = model.eval().cuda()
+
+UPLOAD_FOLDER = "./uploads"
+RESULTS_FOLDER = "./results"
+
+for folder in [UPLOAD_FOLDER, RESULTS_FOLDER]:
+    if not os.path.exists(folder):
+        os.makedirs(folder)
+
+def image_to_base64(image):
+    buffered = io.BytesIO()
+    image.save(buffered, format="PNG")
+    return base64.b64encode(buffered.getvalue()).decode()
+
+@spaces.GPU
+def run_GOT(image, got_mode, fine_grained_mode="", ocr_color="", ocr_box=""):
+    unique_id = str(uuid.uuid4())
+    image_path = os.path.join(UPLOAD_FOLDER, f"{unique_id}.png")
+    result_path = os.path.join(RESULTS_FOLDER, f"{unique_id}.html")
+    
+    shutil.copy(image, image_path)
+    
     try:
-        if image is None:
-            return "No image captured. Please try again."
-
-        # Verify Tesseract executable is accessible
-        try:
-            subprocess.run([pytesseract.pytesseract.tesseract_cmd, "--version"],
-                          check=True, capture_output=True, text=True)
-        except (subprocess.SubprocessError, FileNotFoundError):
-            return "Tesseract OCR is not installed or not properly configured. Please check installation."
-
-        # Image preprocessing for better OCR
-        import cv2
-        import numpy as np
-
-        # Convert PIL image to OpenCV format
-        img_cv = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
-
-        # Convert to grayscale
-        gray = cv2.cvtColor(img_cv, cv2.COLOR_BGR2GRAY)
-
-        # Apply thresholding to get black and white image
-        _, binary = cv2.threshold(gray, 150, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
-
-        # Noise removal
-        kernel = np.ones((1, 1), np.uint8)
-        binary = cv2.morphologyEx(binary, cv2.MORPH_OPEN, kernel)
-
-        # Dilate to connect text
-        binary = cv2.dilate(binary, kernel, iterations=1)
-
-        # Convert back to PIL image for tesseract
-        binary_pil = Image.fromarray(cv2.bitwise_not(binary))
-
-        # Run OCR with improved configuration
-        custom_config = r'--oem 3 --psm 6 -l eng'
-        text = pytesseract.image_to_string(binary_pil, config=custom_config)
-
-        if not text.strip():
-            # Try original image as fallback
-            text = pytesseract.image_to_string(image, config=custom_config)
-
-        if not text.strip():
-            return "No text could be extracted. Ensure image is clear and readable."
-
-        return text
+        if got_mode == "plain texts OCR":
+            res = model.chat(tokenizer, image_path, ocr_type='ocr')
+            return res, None
+        elif got_mode == "format texts OCR":
+            res = model.chat(tokenizer, image_path, ocr_type='format', render=True, save_render_file=result_path)
+        elif got_mode == "plain multi-crop OCR":
+            res = model.chat_crop(tokenizer, image_path, ocr_type='ocr')
+            return res, None
+        elif got_mode == "format multi-crop OCR":
+            res = model.chat_crop(tokenizer, image_path, ocr_type='format', render=True, save_render_file=result_path)
+        elif got_mode == "plain fine-grained OCR":
+            res = model.chat(tokenizer, image_path, ocr_type='ocr', ocr_box=ocr_box, ocr_color=ocr_color)
+            return res, None
+        elif got_mode == "format fine-grained OCR":
+            res = model.chat(tokenizer, image_path, ocr_type='format', ocr_box=ocr_box, ocr_color=ocr_color, render=True, save_render_file=result_path)
+
+        # res_markdown = f"$$ {res} $$"
+        res_markdown = res
+
+        if "format" in got_mode and os.path.exists(result_path):
+            with open(result_path, 'r') as f:
+                html_content = f.read()
+            encoded_html = base64.b64encode(html_content.encode('utf-8')).decode('utf-8')
+            iframe_src = f"data:text/html;base64,{encoded_html}"
+            iframe = f'<iframe src="{iframe_src}" width="100%" height="600px"></iframe>'
+            download_link = f'<a href="data:text/html;base64,{encoded_html}" download="result_{unique_id}.html">Download Full Result</a>'
+            return res_markdown, f"{download_link}<br>{iframe}"
+        else:
+            return res_markdown, None
     except Exception as e:
-        return f"Error extracting text: {str(e)}"
+        return f"Error: {str(e)}", None
+    finally:
+        if os.path.exists(image_path):
+            os.remove(image_path)
+
+def task_update(task):
+    if "fine-grained" in task:
+        return [
+            gr.update(visible=True),
+            gr.update(visible=False),
+            gr.update(visible=False),
+        ]
+    else:
+        return [
+            gr.update(visible=False),
+            gr.update(visible=False),
+            gr.update(visible=False),
+        ]
+
+def fine_grained_update(task):
+    if task == "box":
+        return [
+            gr.update(visible=False, value = ""),
+            gr.update(visible=True),
+        ]
+    elif task == 'color':
+        return [
+            gr.update(visible=True),
+            gr.update(visible=False, value = ""),
+        ]
 
 # Function to parse ingredients from text
 def parse_ingredients(text):