Smart_Bite_AI / app.py
Amandeep01's picture
Update app.py
7943559 verified
import gradio as gr
import numpy as np
import pytesseract
from PIL import Image, ImageEnhance, ImageFilter
import pandas as pd
import re
import json
import os
import random
import datetime
import matplotlib.pyplot as plt
import seaborn as sns
import cv2
import requests
from io import BytesIO
# Load nutrition database with expanded items
def load_nutrition_data():
# Enhanced food database with more items and categories
food_data = {
# Fast food and restaurant items
"pizza": {"calories": 285, "fat": 10, "carbs": 36, "protein": 12, "category": "junk"},
"burger": {"calories": 354, "fat": 17, "carbs": 40, "protein": 15, "category": "junk"},
"cheeseburger": {"calories": 400, "fat": 20, "carbs": 40, "protein": 15, "category": "junk"},
"hamburger": {"calories": 350, "fat": 15, "carbs": 40, "protein": 15, "category": "junk"},
"fries": {"calories": 312, "fat": 15, "carbs": 41, "protein": 3, "category": "junk"},
"french fries": {"calories": 312, "fat": 15, "carbs": 41, "protein": 3, "category": "junk"},
"salad": {"calories": 100, "fat": 7, "carbs": 5, "protein": 2, "category": "healthy"},
"caesar salad": {"calories": 150, "fat": 10, "carbs": 5, "protein": 3, "category": "healthy"},
"garden salad": {"calories": 80, "fat": 5, "carbs": 5, "protein": 2, "category": "healthy"},
"soda": {"calories": 140, "fat": 0, "carbs": 39, "protein": 0, "category": "junk"},
"coke": {"calories": 140, "fat": 0, "carbs": 39, "protein": 0, "category": "junk"},
"pepsi": {"calories": 150, "fat": 0, "carbs": 41, "protein": 0, "category": "junk"},
"sprite": {"calories": 140, "fat": 0, "carbs": 38, "protein": 0, "category": "junk"},
"cola": {"calories": 140, "fat": 0, "carbs": 39, "protein": 0, "category": "junk"},
"diet coke": {"calories": 0, "fat": 0, "carbs": 0, "protein": 0, "category": "neutral"},
"juice": {"calories": 110, "fat": 0, "carbs": 26, "protein": 0, "category": "neutral"},
"orange juice": {"calories": 110, "fat": 0, "carbs": 26, "protein": 0, "category": "neutral"},
"apple juice": {"calories": 115, "fat": 0, "carbs": 28, "protein": 0, "category": "neutral"},
"water": {"calories": 0, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
"sparkling water": {"calories": 0, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
"pasta": {"calories": 200, "fat": 2, "carbs": 42, "protein": 7, "category": "neutral"},
"spaghetti": {"calories": 220, "fat": 2, "carbs": 43, "protein": 8, "category": "neutral"},
"pasta carbonara": {"calories": 380, "fat": 18, "carbs": 43, "protein": 14, "category": "neutral"},
"fettuccine": {"calories": 220, "fat": 2, "carbs": 43, "protein": 8, "category": "neutral"},
"lasagna": {"calories": 360, "fat": 12, "carbs": 37, "protein": 25, "category": "neutral"},
"mac and cheese": {"calories": 350, "fat": 15, "carbs": 45, "protein": 15, "category": "neutral"},
"macaroni": {"calories": 200, "fat": 2, "carbs": 42, "protein": 7, "category": "neutral"},
"steak": {"calories": 300, "fat": 15, "carbs": 0, "protein": 30, "category": "protein"},
"ribeye": {"calories": 330, "fat": 25, "carbs": 0, "protein": 30, "category": "protein"},
"filet mignon": {"calories": 320, "fat": 20, "carbs": 0, "protein": 35, "category": "protein"},
"sirloin": {"calories": 270, "fat": 12, "carbs": 0, "protein": 32, "category": "protein"},
"chicken": {"calories": 220, "fat": 8, "carbs": 0, "protein": 40, "category": "protein"},
"chicken wings": {"calories": 350, "fat": 18, "carbs": 5, "protein": 33, "category": "protein"},
"chicken tenders": {"calories": 380, "fat": 20, "carbs": 20, "protein": 30, "category": "protein"},
"grilled chicken": {"calories": 220, "fat": 8, "carbs": 0, "protein": 40, "category": "protein"},
"fried chicken": {"calories": 320, "fat": 16, "carbs": 12, "protein": 28, "category": "protein"},
"fish": {"calories": 180, "fat": 5, "carbs": 0, "protein": 30, "category": "healthy"},
"salmon": {"calories": 200, "fat": 10, "carbs": 0, "protein": 25, "category": "healthy"},
"tuna": {"calories": 160, "fat": 3, "carbs": 0, "protein": 33, "category": "healthy"},
"cod": {"calories": 150, "fat": 2, "carbs": 0, "protein": 28, "category": "healthy"},
"rice": {"calories": 130, "fat": 0, "carbs": 28, "protein": 3, "category": "neutral"},
"brown rice": {"calories": 110, "fat": 1, "carbs": 22, "protein": 3, "category": "healthy"},
"white rice": {"calories": 130, "fat": 0, "carbs": 28, "protein": 3, "category": "neutral"},
"fried rice": {"calories": 230, "fat": 10, "carbs": 28, "protein": 8, "category": "neutral"},
#Indian Food Items
"butter chicken": {"calories": 450, "fat": 28, "carbs": 14, "protein": 32, "category": "protein"},
"chole bhature": {"calories": 550, "fat": 30, "carbs": 50, "protein": 14, "category": "junk"},
"dal makhani": {"calories": 320, "fat": 18, "carbs": 24, "protein": 14, "category": "neutral"},
"rajma chawal": {"calories": 410, "fat": 10, "carbs": 60, "protein": 15, "category": "neutral"},
"paneer butter masala": {"calories": 430, "fat": 30, "carbs": 20, "protein": 18, "category": "protein"},
"tandoori chicken": {"calories": 290, "fat": 13, "carbs": 4, "protein": 35, "category": "protein"},
"biryani": {"calories": 480, "fat": 20, "carbs": 55, "protein": 18, "category": "neutral"},
"veg biryani": {"calories": 400, "fat": 15, "carbs": 50, "protein": 10, "category": "neutral"},
"chicken biryani": {"calories": 500, "fat": 20, "carbs": 55, "protein": 22, "category": "protein"},
"aloo paratha": {"calories": 300, "fat": 12, "carbs": 40, "protein": 7, "category": "neutral"},
"samosa": {"calories": 260, "fat": 15, "carbs": 25, "protein": 5, "category": "junk"},
"masala dosa": {"calories": 390, "fat": 15, "carbs": 50, "protein": 8, "category": "neutral"},
"idli sambar": {"calories": 270, "fat": 6, "carbs": 45, "protein": 10, "category": "healthy"},
"pav bhaji": {"calories": 420, "fat": 22, "carbs": 40, "protein": 8, "category": "junk"},
"poha": {"calories": 250, "fat": 10, "carbs": 35, "protein": 5, "category": "healthy"},
"upma": {"calories": 240, "fat": 8, "carbs": 34, "protein": 6, "category": "healthy"},
"chana masala": {"calories": 280, "fat": 12, "carbs": 30, "protein": 12, "category": "protein"},
"fish curry": {"calories": 310, "fat": 18, "carbs": 12, "protein": 25, "category": "protein"},
"mutton curry": {"calories": 500, "fat": 35, "carbs": 10, "protein": 30, "category": "protein"},
"kadai paneer": {"calories": 380, "fat": 25, "carbs": 18, "protein": 15, "category": "protein"},
"malai kofta": {"calories": 440, "fat": 30, "carbs": 25, "protein": 12, "category": "junk"},
"dal tadka": {"calories": 280, "fat": 10, "carbs": 30, "protein": 12, "category": "healthy"},
"aloo methi": {"calories": 200, "fat": 10, "carbs": 18, "protein": 6, "category": "healthy"},
"phulka": {"calories": 90, "fat": 2, "carbs": 18, "protein": 3, "category": "healthy"},
"butter naan": {"calories": 300, "fat": 10, "carbs": 40, "protein": 6, "category": "junk"},
"tandoori roti": {"calories": 120, "fat": 2, "carbs": 22, "protein": 4, "category": "healthy"},
"mineral water": {"calories": 0, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
"Chicken Tikka Masala": {"calories": 480, "fat": 28, "carbs": 16, "protein": 35, "category": "protein"},
"mutton biryani": {"calories": 550, "fat": 25, "carbs": 50, "protein": 30, "category": "protein"},
"garlic naan": {"calories": 160, "fat": 5, "carbs": 28, "protein": 4, "category": "neutral"},
"garlic naan (2 pcs)": {"calories": 320, "fat": 10, "carbs": 56, "protein": 8, "category": "neutral"},
"butter naan (2 pcs)": {"calories": 600, "fat": 20, "carbs": 80, "protein": 12, "category": "junk"},
"jeera rice": {"calories": 210, "fat": 5, "carbs": 35, "protein": 4, "category": "neutral"},
"papadum": {"calories": 40, "fat": 2, "carbs": 4, "protein": 1, "category": "neutral"},
"papadum": {"calories": 160, "fat": 8, "carbs": 16, "protein": 4, "category": "neutral"},
"mixed raita": {"calories": 100, "fat": 5, "carbs": 10, "protein": 3, "category": "healthy"},
"gulab jamun": {"calories": 150, "fat": 7, "carbs": 25, "protein": 3, "category": "junk"},
"gulab jamun 4 pcs": {"calories": 600, "fat": 28, "carbs": 100, "protein": 12, "category": "junk"},
"masala chai": {"calories": 80, "fat": 3, "carbs": 10, "protein": 2, "category": "neutral"},
"masala chai 4 cups": {"calories": 320, "fat": 12, "carbs": 40, "protein": 8, "category": "neutral"},
"pongal": {"calories": 320, "fat": 12, "carbs": 40, "protein": 7, "category": "neutral"},
"medu vadai": {"calories": 150, "fat": 8, "carbs": 15, "protein": 4, "category": "junk"},
"sambhar idly": {"calories": 270, "fat": 6, "carbs": 40, "protein": 10, "category": "healthy"},
"poori": {"calories": 150, "fat": 8, "carbs": 18, "protein": 3, "category": "junk"},
"ghee roast": {"calories": 450, "fat": 25, "carbs": 45, "protein": 7, "category": "junk"},
"tea": {"calories": 50, "fat": 2, "carbs": 5, "protein": 1, "category": "neutral"},
"dal makhani": {"calories": 320, "fat": 18, "carbs": 24, "protein": 14, "category": "neutral"},
"mixed raita": {"calories": 100, "fat": 5, "carbs": 10, "protein": 3, "category": "healthy"},
"gulab jamun": {"calories": 150, "fat": 7, "carbs": 25, "protein": 3, "category": "junk"},
"masala chai": {"calories": 80, "fat": 3, "carbs": 10, "protein": 2, "category": "neutral"},
"mutton biryani": {"calories": 550, "fat": 25, "carbs": 50, "protein": 30, "category": "protein"},
# Drinks
"beer": {"calories": 154, "fat": 0, "carbs": 13, "protein": 1, "category": "junk"},
"wine": {"calories": 125, "fat": 0, "carbs": 4, "protein": 0, "category": "neutral"},
"red wine": {"calories": 125, "fat": 0, "carbs": 4, "protein": 0, "category": "neutral"},
"white wine": {"calories": 120, "fat": 0, "carbs": 4, "protein": 0, "category": "neutral"},
"cocktail": {"calories": 180, "fat": 0, "carbs": 20, "protein": 0, "category": "junk"},
"margarita": {"calories": 200, "fat": 0, "carbs": 25, "protein": 0, "category": "junk"},
"daiquiri": {"calories": 180, "fat": 0, "carbs": 20, "protein": 0, "category": "junk"},
"mojito": {"calories": 160, "fat": 0, "carbs": 18, "protein": 0, "category": "junk"},
"martini": {"calories": 120, "fat": 0, "carbs": 3, "protein": 0, "category": "neutral"},
"coffee": {"calories": 5, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
"latte": {"calories": 120, "fat": 4, "carbs": 10, "protein": 8, "category": "neutral"},
"cappuccino": {"calories": 110, "fat": 4, "carbs": 8, "protein": 6, "category": "neutral"},
"espresso": {"calories": 5, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
# Desserts
"ice cream": {"calories": 207, "fat": 11, "carbs": 24, "protein": 4, "category": "junk"},
"cake": {"calories": 350, "fat": 18, "carbs": 45, "protein": 4, "category": "junk"},
"chocolate cake": {"calories": 370, "fat": 19, "carbs": 48, "protein": 5, "category": "junk"},
"cheesecake": {"calories": 400, "fat": 25, "carbs": 35, "protein": 7, "category": "junk"},
"tiramisu": {"calories": 380, "fat": 20, "carbs": 40, "protein": 5, "category": "junk"},
"brownie": {"calories": 300, "fat": 15, "carbs": 40, "protein": 3, "category": "junk"},
"cookie": {"calories": 180, "fat": 9, "carbs": 22, "protein": 2, "category": "junk"},
"chocolate": {"calories": 200, "fat": 12, "carbs": 20, "protein": 2, "category": "junk"},
"pie": {"calories": 300, "fat": 14, "carbs": 38, "protein": 3, "category": "junk"},
"apple pie": {"calories": 290, "fat": 14, "carbs": 40, "protein": 3, "category": "junk"},
"pudding": {"calories": 150, "fat": 4, "carbs": 25, "protein": 3, "category": "junk"},
# Other common items
"sandwich": {"calories": 250, "fat": 8, "carbs": 30, "protein": 15, "category": "neutral"},
"wrap": {"calories": 220, "fat": 5, "carbs": 30, "protein": 13, "category": "neutral"},
"soup": {"calories": 120, "fat": 3, "carbs": 12, "protein": 10, "category": "healthy"},
"bread": {"calories": 80, "fat": 1, "carbs": 15, "protein": 3, "category": "neutral"},
"garlic bread": {"calories": 150, "fat": 6, "carbs": 18, "protein": 4, "category": "neutral"},
"roll": {"calories": 80, "fat": 1, "carbs": 15, "protein": 3, "category": "neutral"},
"milkshake": {"calories": 300, "fat": 10, "carbs": 50, "protein": 9, "category": "junk"},
"dessert": {"calories": 280, "fat": 14, "carbs": 35, "protein": 5, "category": "junk"},
"smoothie": {"calories": 170, "fat": 2, "carbs": 35, "protein": 5, "category": "neutral"},
"tea": {"calories": 2, "fat": 0, "carbs": 0, "protein": 0, "category": "healthy"},
"appetizer": {"calories": 200, "fat": 12, "carbs": 15, "protein": 8, "category": "neutral"},
"noodles": {"calories": 190, "fat": 2, "carbs": 40, "protein": 7, "category": "neutral"},
"taco": {"calories": 210, "fat": 10, "carbs": 22, "protein": 12, "category": "neutral"},
"burrito": {"calories": 350, "fat": 12, "carbs": 50, "protein": 15, "category": "neutral"},
"nachos": {"calories": 600, "fat": 35, "carbs": 58, "protein": 20, "category": "junk"},
"fajitas": {"calories": 290, "fat": 10, "carbs": 30, "protein": 25, "category": "neutral"},
"quesadilla": {"calories": 400, "fat": 22, "carbs": 35, "protein": 18, "category": "neutral"},
"eggs": {"calories": 140, "fat": 10, "carbs": 1, "protein": 12, "category": "protein"},
"omelette": {"calories": 220, "fat": 16, "carbs": 2, "protein": 16, "category": "protein"},
"pancakes": {"calories": 380, "fat": 12, "carbs": 60, "protein": 10, "category": "neutral"},
"waffles": {"calories": 370, "fat": 14, "carbs": 55, "protein": 8, "category": "neutral"},
"toast": {"calories": 80, "fat": 1, "carbs": 15, "protein": 3, "category": "neutral"},
"muffin": {"calories": 210, "fat": 10, "carbs": 30, "protein": 3, "category": "junk"},
"croissant": {"calories": 230, "fat": 12, "carbs": 26, "protein": 5, "category": "neutral"},
"doughnut": {"calories": 250, "fat": 12, "carbs": 30, "protein": 4, "category": "junk"},
"donut": {"calories": 250, "fat": 12, "carbs": 30, "protein": 4, "category": "junk"},
"bagel": {"calories": 245, "fat": 1, "carbs": 48, "protein": 10, "category": "neutral"},
"scone": {"calories": 230, "fat": 12, "carbs": 28, "protein": 4, "category": "neutral"},
# Side dishes
"onion rings": {"calories": 320, "fat": 18, "carbs": 35, "protein": 5, "category": "junk"},
"mashed potatoes": {"calories": 150, "fat": 4, "carbs": 25, "protein": 3, "category": "neutral"},
"baked potato": {"calories": 130, "fat": 0, "carbs": 30, "protein": 3, "category": "neutral"},
"coleslaw": {"calories": 120, "fat": 8, "carbs": 10, "protein": 1, "category": "neutral"},
"corn": {"calories": 90, "fat": 1, "carbs": 20, "protein": 3, "category": "healthy"},
"broccoli": {"calories": 40, "fat": 0, "carbs": 8, "protein": 4, "category": "healthy"},
"veggies": {"calories": 50, "fat": 0, "carbs": 10, "protein": 2, "category": "healthy"},
"vegetables": {"calories": 50, "fat": 0, "carbs": 10, "protein": 2, "category": "healthy"},
"chips": {"calories": 300, "fat": 15, "carbs": 35, "protein": 3, "category": "junk"},
}
return food_data
# Load nutrition database
nutrition_data = load_nutrition_data()
# Load motivational quotes based on health score ranges
def load_motivational_quotes():
quotes = {
"excellent": [
"You're making excellent food choices! Your body thanks you for the premium fuel.",
"Fantastic choices! You're investing in your long-term health with every bite.",
"Your healthy eating habits today are building your stronger body for tomorrow.",
"Impressive meal choices! You're mastering the art of nutritious eating.",
"You're a nutrition champion! These balanced choices will energize your day."
],
"good": [
"Good job balancing nutrition! Small improvements can take you to the next level.",
"You're on the right track with your food choices. Keep building those healthy habits!",
"Nice work choosing a fairly balanced meal. Your body appreciates the consideration.",
"Your meal choices show you care about your health. Keep that momentum going!",
"Good balance of nutrients in this meal. Remember: consistency is key to health."
],
"moderate": [
"This meal has some nutritional bright spots. Consider adding more protein next time.",
"Balance is a journey. Try adding more vegetables to your next meal.",
"Everyone indulges sometimes. Tomorrow is a new opportunity for nourishing choices.",
"Consider this meal a starting point. Small improvements add up to big health benefits.",
"Moderation is key. Try balancing this meal with healthier choices later today."
],
"poor": [
"Your body deserves premium fuel. Consider more nutrient-dense options next time.",
"One meal doesn't define your health journey. Your next choice can be a healthier one.",
"We all have indulgences. Balance this meal with nutritious choices for your next one.",
"Small steps lead to big changes. Consider adding vegetables to your next meal.",
"Remember: food is fuel. Choose options that will energize rather than drain you."
]
}
return quotes
# Initialize motivational quotes
motivational_quotes = load_motivational_quotes()
# Helper function to preprocess the image for better OCR results
def preprocess_image(image):
# Convert to numpy array if needed
if not isinstance(image, np.ndarray):
image = np.array(image)
try:
# Ensure the image is in RGB format (3 channels)
if len(image.shape) == 2: # Grayscale
image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
elif len(image.shape) == 3 and image.shape[2] == 4: # RGBA
image = cv2.cvtColor(image, cv2.COLOR_RGBA2RGB)
# Convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
# Apply multiple preprocessing techniques and keep the best result
results = []
# Technique 1: Adaptive thresholding
thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, 11, 2)
results.append(thresh)
# Technique 2: Otsu's thresholding after Gaussian filtering
blur = cv2.GaussianBlur(gray, (5, 5), 0)
_, otsu = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
results.append(otsu)
# Technique 3: Histogram equalization
equalized = cv2.equalizeHist(gray)
results.append(equalized)
# Technique 4: Original grayscale
results.append(gray)
# Convert all results to PIL images
pil_images = [Image.fromarray(img) for img in results]
return pil_images
except Exception as e:
print(f"Error preprocessing image: {e}")
# If preprocessing fails, return the original image as a list
return [Image.fromarray(image) if isinstance(image, np.ndarray) else image]
# OCR function to extract text from bill image with enhanced image processing
def extract_text_from_image(image):
try:
# If image is a URL, download it
if isinstance(image, str) and (image.startswith('http://') or image.startswith('https://')):
response = requests.get(image)
img = Image.open(BytesIO(response.content))
else:
img = Image.fromarray(image) if isinstance(image, np.ndarray) else image
# Create a copy for display
display_img = img.copy() if hasattr(img, 'copy') else img
# Preprocess the image to get multiple versions
preprocessed_images = preprocess_image(img)
# Try OCR on each preprocessed image
best_text = ""
# Custom configs to try
configs = [
r'--oem 3 --psm 6 -l eng', # Assume a single uniform block of text
r'--oem 3 --psm 4 -l eng', # Assume a single column of text
r'--oem 3 --psm 3 -l eng', # Fully automatic page segmentation
r'--oem 3 --psm 11 -l eng', # Sparse text - no specific structure
r'--oem 3 --psm 12 -l eng', # Sparse text with OSD
]
for img_version in preprocessed_images:
for config in configs:
try:
text = pytesseract.image_to_string(img_version, config=config)
# Keep the longest text as it likely contains more information
if len(text.strip()) > len(best_text.strip()):
best_text = text
except Exception as e:
print(f"OCR error with specific config: {str(e)}")
continue
# If all attempts failed or returned very little text
if len(best_text.strip()) < 10:
# Try one last attempt with default settings
try:
best_text = pytesseract.image_to_string(img)
except Exception as e:
print(f"Final OCR attempt error: {str(e)}")
# Debug output
print(f"OCR extracted text of length: {len(best_text)}")
return best_text
except Exception as e:
print(f"Error extracting text: {str(e)}")
return f"Error extracting text: {str(e)}"
# Extract food items from the OCR text with improved pattern recognition
def extract_food_items(text):
# Improved algorithm to detect food items in bill text
lines = text.split('\n')
food_items = []
# Debug info
print(f"Processing {len(lines)} lines of text")
# Clean and normalize all lines first
cleaned_lines = []
for line in lines:
# Remove common non-food text
line = re.sub(r'thank you|receipt|invoice|order|table|server', '', line.lower(), flags=re.IGNORECASE)
cleaned_lines.append(line.strip())
# Regular patterns for food items in bills
# More comprehensive price pattern to catch various formats
price_pattern = r'(\$?\d+\.\d{2}|\$?\d+\,\d{2}|\$?\d+)'
for line in cleaned_lines:
if not line:
continue
# Skip lines that look like totals or headers
skip_keywords = [
'total', 'subtotal', 'tax', 'gratuity', 'tip', 'service', 'amount', 'due', 'change',
'cash', 'credit', 'card', 'payment', 'date', 'time', 'check', 'table',
'guest', 'invoice', 'receipt', 'bill', 'order', 'tel', 'phone', 'address',
'thank you', 'restaurant', 'cafe', 'bar', 'grill', 'kitchen', 'www', 'http'
]
if any(keyword in line.lower() for keyword in skip_keywords):
continue
# Debug line
print(f"Processing line: '{line}'")
# If line contains a price, extract the item name (everything before the price)
if re.search(price_pattern, line):
# Split based on number patterns (likely price)
item_parts = re.split(price_pattern, line)
if item_parts and len(item_parts) > 1:
item_match = item_parts[0].strip()
if item_match and len(item_match) > 1: # Ensure it's not just whitespace
# Clean up the item name (remove quantities, etc.)
cleaned_item = re.sub(r'^\d+\s*[xX]?\s*', '', item_match) # Remove quantities like "2 x" or "2"
cleaned_item = re.sub(r'\d+\s*oz\s*', '', cleaned_item) # Remove sizes like "12oz"
cleaned_item = re.sub(r'\(\w+\)', '', cleaned_item) # Remove parentheses
# Filter out very short items that are likely not food
if len(cleaned_item.strip()) > 2:
food_items.append(cleaned_item.strip().lower())
print(f"Found item with price: '{cleaned_item.strip().lower()}'")
# If not enough items found, try alternate methods
if len(food_items) < 2:
# Look for menu-like patterns
for line in cleaned_lines:
# Try to find numbered items (e.g., "1. Burger" or "#1 Burger")
numbered_pattern = r'(?:^|\s)(?:\d+\.|\#\d+)\s+(.+?)(?:\s+\$|\s+\d|\s*$)'
match = re.search(numbered_pattern, line)
if match:
item = match.group(1).strip().lower()
if len(item) > 2 and item not in food_items:
food_items.append(item)
print(f"Found numbered item: '{item}'")
# Simple heuristic: look for capitalized words that might be menu items
# This is a fallback when we're struggling to find items
if len(line) > 3 and not any(char.isdigit() for char in line) and not any(skip in line for skip in skip_keywords):
potential_item = re.sub(r'\W+', ' ', line).strip().lower()
# Check if the line contains any known food items
for food in nutrition_data.keys():
if food in potential_item:
if potential_item not in food_items:
food_items.append(potential_item)
print(f"Found potential food item: '{potential_item}'")
break
# If we still have no items, use a more aggressive approach to find any words
# that match our food database
if len(food_items) < 2:
print("Using aggressive food item detection...")
# Flatten all text and clean it
all_text = ' '.join(cleaned_lines).lower()
# Filter out non-alphanumeric characters
all_text = re.sub(r'[^\w\s]', ' ', all_text)
# Get all words
words = all_text.split()
# Look for any word or pair of words that matches our food database
for i in range(len(words)):
# Single word match
if words[i] in nutrition_data:
food_items.append(words[i])
print(f"Found direct food match: '{words[i]}'")
# Two-word match
if i < len(words) - 1:
two_words = words[i] + ' ' + words[i+1]
if two_words in nutrition_data:
food_items.append(two_words)
print(f"Found direct two-word food match: '{two_words}'")
# If we've exhausted all options but still have no items, try to find words
# that are similar to our food database
if len(food_items) < 2:
print("Using similarity-based food item detection...")
all_text = ' '.join(cleaned_lines).lower()
words = re.findall(r'\b[a-z]{3,}\b', all_text) # Find all words with at least 3 letters
for word in words:
# Skip very common words
if word in ['the', 'and', 'for', 'with', 'that', 'have', 'this', 'from']:
continue
# Check if the word is a substring of any food in our database
for food in nutrition_data.keys():
if word in food:
food_items.append(food)
print(f"Found similar food item: '{food}' from '{word}'")
break
# Remove duplicates and limit to reasonable number
food_items = list(set(food_items))[:10]
print(f"Final food items extracted: {food_items}")
return food_items
# Match extracted food items to our nutrition database with improved fuzzy matching
def match_food_to_nutrition(food_items):
matched_items = []
for item in food_items:
# Direct match
if item in nutrition_data:
matched_items.append({"name": item, "nutrition": nutrition_data[item]})
continue
# Improved matching logic - word-based matching and ngram similarity
best_match = None
max_score = 0
# Split the item into words for better matching
item_words = set(item.split())
for db_food in nutrition_data:
# Calculate word overlap
db_food_words = set(db_food.split())
if item_words and db_food_words:
overlap = len(item_words.intersection(db_food_words))
score = overlap / max(len(item_words), len(db_food_words))
# Boost score if one string contains the other
if db_food in item or item in db_food:
score += 0.3
if score > max_score:
max_score = score
best_match = db_food
# Only match if the score is reasonably high
if best_match and max_score > 0.3:
matched_items.append({"name": item, "matched_as": best_match, "nutrition": nutrition_data[best_match]})
# Remove duplicates (based on matched_as)
unique_matches = []
seen_matches = set()
for item in matched_items:
match_key = item.get("matched_as", item["name"])
if match_key not in seen_matches:
unique_matches.append(item)
seen_matches.add(match_key)
return unique_matches
# Calculate nutritional totals and health
# Calculate nutritional totals and health score
def calculate_meal_health(matched_items):
if not matched_items:
return None, None, "No food items detected"
# Calculate total nutrition
total_calories = sum(item["nutrition"]["calories"] for item in matched_items)
total_fat = sum(item["nutrition"]["fat"] for item in matched_items)
total_carbs = sum(item["nutrition"]["carbs"] for item in matched_items)
total_protein = sum(item["nutrition"]["protein"] for item in matched_items)
# Count items by category
category_counts = {"healthy": 0, "neutral": 0, "protein": 0, "junk": 0}
for item in matched_items:
category = item["nutrition"]["category"]
category_counts[category] = category_counts.get(category, 0) + 1
# Calculate health score (0-100)
total_items = len(matched_items)
health_score = 0
# Point system:
# - Healthy items: +25 points each
# - Protein items: +15 points each
# - Neutral items: +5 points each
# - Junk items: -10 points each
# Base score of 50
health_score = 50
health_score += category_counts["healthy"] * 25
health_score += category_counts["protein"] * 15
health_score += category_counts["neutral"] * 5
health_score -= category_counts["junk"] * 10
# Adjust based on macros
if total_calories > 0:
# Protein is good
protein_ratio = (total_protein * 4) / total_calories
if protein_ratio > 0.25: # >25% protein is good
health_score += 10
# Too much fat is not ideal
fat_ratio = (total_fat * 9) / total_calories
if fat_ratio > 0.4: # >40% calories from fat
health_score -= 10
# Clamp score between 0-100
health_score = max(0, min(100, health_score))
# Determine feedback category
if health_score >= 80:
category = "excellent"
elif health_score >= 60:
category = "good"
elif health_score >= 40:
category = "moderate"
else:
category = "poor"
# Get a random motivational quote for the category
quote = random.choice(motivational_quotes[category])
# Create nutrition data
nutrition_data = {
"calories": total_calories,
"fat": total_fat,
"carbs": total_carbs,
"protein": total_protein,
"health_score": health_score,
"category": category,
"message": quote
}
# Create summary
dominant_macro = ""
if total_calories > 0:
fat_percentage = (total_fat * 9) / total_calories * 100
carbs_percentage = (total_carbs * 4) / total_calories * 100
protein_percentage = (total_protein * 4) / total_calories * 100
if max(fat_percentage, carbs_percentage, protein_percentage) == fat_percentage:
dominant_macro = "fat"
elif max(fat_percentage, carbs_percentage, protein_percentage) == carbs_percentage:
dominant_macro = "carbs"
else:
dominant_macro = "protein"
summary = f"You consumed approximately {total_calories} calories — mostly {dominant_macro}."
if health_score >= 70:
summary += " Great choices today!"
elif health_score >= 50:
summary += " Consider more balanced options next time."
else:
summary += " Try to make healthier choices next time."
return nutrition_data, summary, ""
# Generate detailed analysis with visualization
def generate_analysis(matched_items, nutrition_data):
if not matched_items or not nutrition_data:
return None
# Create DataFrame for the items
items_data = []
for item in matched_items:
name = item["name"]
if "matched_as" in item:
name = f"{name} (matched as {item['matched_as']})"
items_data.append({
"Item": name,
"Calories": item["nutrition"]["calories"],
"Fat (g)": item["nutrition"]["fat"],
"Carbs (g)": item["nutrition"]["carbs"],
"Protein (g)": item["nutrition"]["protein"],
"Category": item["nutrition"]["category"].capitalize()
})
df = pd.DataFrame(items_data)
# Get the current date and time
now = datetime.datetime.now()
date_str = now.strftime("%Y-%m-%d")
time_str = now.strftime("%H:%M:%S")
# Create visualization plots
fig, axs = plt.subplots(2, 2, figsize=(12, 10))
# Plot 1: Calories by item (horizontal bar)
df_sorted = df.sort_values('Calories', ascending=True)
sns.barplot(x='Calories', y='Item', data=df_sorted, ax=axs[0, 0], palette='viridis')
axs[0, 0].set_title('Calories by Item')
axs[0, 0].set_xlabel('Calories')
axs[0, 0].set_ylabel('Food Item')
# Plot 2: Macronutrient breakdown (pie chart)
total_calories = nutrition_data["calories"]
if total_calories > 0:
fat_cals = nutrition_data["fat"] * 9
carb_cals = nutrition_data["carbs"] * 4
protein_cals = nutrition_data["protein"] * 4
macro_data = [fat_cals, carb_cals, protein_cals]
macro_labels = [f'Fat ({fat_cals:.0f} cal)', f'Carbs ({carb_cals:.0f} cal)', f'Protein ({protein_cals:.0f} cal)']
colors = ['#FF9999', '#66B2FF', '#99FF99']
axs[0, 1].pie(macro_data, labels=macro_labels, colors=colors, autopct='%1.1f%%', startangle=90)
axs[0, 1].set_title('Calorie Sources')
else:
axs[0, 1].text(0.5, 0.5, 'No calorie data available', ha='center', va='center')
axs[0, 1].axis('off')
# Plot 3: Health score gauge
health_score = nutrition_data["health_score"]
# Create a gauge chart using a pie chart
size = 0.3
vals = [health_score, 100-health_score]
# Create color based on score
if health_score >= 80:
color = '#00CC66' # Green
elif health_score >= 60:
color = '#CCCC00' # Yellow
elif health_score >= 40:
color = '#FF9900' # Orange
else:
color = '#FF3333' # Red
cmap = [color, '#f0f0f0']
axs[1, 0].pie(vals, radius=1, colors=cmap, startangle=90, counterclock=False)
axs[1, 0].pie([1], radius=1-size, colors=['white'])
axs[1, 0].text(0, 0, f"{health_score:.0f}", fontsize=32, ha='center', va='center')
axs[1, 0].text(0, -0.2, "Health Score", fontsize=12, ha='center', va='center')
axs[1, 0].set_title('Meal Health Score')
# Plot 4: Food category breakdown
category_counts = df['Category'].value_counts()
sns.barplot(x=category_counts.index, y=category_counts.values, ax=axs[1, 1], palette='viridis')
axs[1, 1].set_title('Food Categories')
axs[1, 1].set_xlabel('Category')
axs[1, 1].set_ylabel('Count')
plt.tight_layout()
# Save the plots to a file
analysis_img_path = "analysis_temp.png"
plt.savefig(analysis_img_path, dpi=150, bbox_inches='tight')
plt.close()
# Create a table of the analyzed items
items_table = df.to_html(index=False, classes='table table-striped')
# Create analysis summary
total_fat = nutrition_data["fat"]
total_carbs = nutrition_data["carbs"]
total_protein = nutrition_data["protein"]
analysis_summary = f"""
<h2>Meal Nutrition Analysis</h2>
<p><strong>Date:</strong> {date_str} &nbsp; <strong>Time:</strong> {time_str}</p>
<h3>Summary</h3>
<p>
Total Calories: <strong>{total_calories:.0f}</strong><br>
Total Fat: <strong>{total_fat:.1f}g</strong> ({(total_fat * 9 / total_calories * 100):.1f}% of calories)<br>
Total Carbs: <strong>{total_carbs:.1f}g</strong> ({(total_carbs * 4 / total_calories * 100):.1f}% of calories)<br>
Total Protein: <strong>{total_protein:.1f}g</strong> ({(total_protein * 4 / total_calories * 100):.1f}% of calories)<br>
Health Score: <strong>{health_score:.0f}/100</strong> ({nutrition_data["category"].capitalize()})
</p>
<h3>Feedback</h3>
<p>{nutrition_data["message"]}</p>
<h3>Analyzed Items</h3>
{items_table}
<h3>Recommendations</h3>
"""
# Add custom recommendations based on the nutritional analysis
if total_protein < 20:
analysis_summary += "<p>✅ <strong>Add more protein</strong> to your meals. Good sources include lean meats, fish, eggs, tofu, or legumes.</p>"
if (total_fat * 9 / total_calories) > 0.4:
analysis_summary += "<p>✅ <strong>Consider reducing fat intake</strong>, especially from fried foods and processed items.</p>"
category_counts_dict = df['Category'].value_counts().to_dict()
junk_count = category_counts_dict.get('Junk', 0)
healthy_count = category_counts_dict.get('Healthy', 0)
if junk_count > healthy_count:
analysis_summary += "<p>✅ <strong>Try to include more fruits and vegetables</strong> in your meals for better nutrition.</p>"
if health_score < 50:
analysis_summary += "<p>✅ <strong>Balance your plate</strong> with 1/2 vegetables, 1/4 protein, and 1/4 whole grains for improved nutrition.</p>"
return analysis_img_path, analysis_summary
# Function to process the bill image with enhanced error handling
def process_bill_image(image):
try:
display_img = None
ocr_text = ""
food_items = []
matched_items = []
nutrition_data = None
summary = ""
error_message = ""
# Process the image if it's valid
if image is not None:
# Extract text using OCR
ocr_text = extract_text_from_image(image)
if ocr_text:
# Extract food items from the OCR text
food_items = extract_food_items(ocr_text)
if food_items:
# Match food items to nutrition database
matched_items = match_food_to_nutrition(food_items)
if matched_items:
# Calculate health score and nutrition data
nutrition_data, summary, error_message = calculate_meal_health(matched_items)
else:
error_message = "No matching food items found in our database. Please try another image."
else:
error_message = "No food items detected. Please try another image or check the image clarity."
else:
error_message = "No text could be extracted from the image. Please try a clearer image."
else:
error_message = "Please upload an image to analyze."
# Generate the food items section
food_items_html = "<p>No food items detected</p>"
if matched_items:
food_items_html = "<ul>"
for item in matched_items:
item_name = item["name"]
if "matched_as" in item:
item_name = f"{item_name} (recognized as {item['matched_as']})"
cals = item["nutrition"]["calories"]
cat = item["nutrition"]["category"].capitalize()
# Choose color based on category
color = "#000000"
if item["nutrition"]["category"] == "healthy":
color = "#007700" # Green
elif item["nutrition"]["category"] == "junk":
color = "#CC0000" # Red
elif item["nutrition"]["category"] == "protein":
color = "#0000CC" # Blue
food_items_html += f'<li style="color:{color};"><strong>{item_name}</strong>: {cals} calories ({cat})</li>'
food_items_html += "</ul>"
# Generate detailed analysis if we have data
analysis_img = None
analysis_html = "<p>No analysis available</p>"
if matched_items and nutrition_data:
try:
analysis_img, analysis_html = generate_analysis(matched_items, nutrition_data)
except Exception as e:
print(f"Error generating analysis: {str(e)}")
analysis_html = f"<p>Error generating analysis: {str(e)}</p>"
# Return the results
return (
ocr_text,
food_items_html,
summary if summary else error_message,
analysis_img if analysis_img else None,
analysis_html
)
except Exception as e:
error_msg = f"An error occurred: {str(e)}"
print(error_msg)
return (
"",
"<p>No food items detected</p>",
error_msg,
None,
"<p>Analysis not available due to an error</p>"
)
# Function to process direct text input (instead of an image)
def process_text_input(text_input):
try:
if not text_input:
return "<p>No text provided</p>", "Please enter some text to analyze", None, "<p>Analysis not available</p>"
# Extract food items from the text
food_items = extract_food_items(text_input)
if not food_items:
return "<p>No food items detected in your text</p>", "No food items found. Try being more specific about what you ate.", None, "<p>Analysis not available</p>"
# Match food items to nutrition database
matched_items = match_food_to_nutrition(food_items)
if not matched_items:
return "<p>No matching food items found in our database</p>", "Your food items couldn't be matched to our database. Try different foods or descriptions.", None, "<p>Analysis not available</p>"
# Calculate health score and nutrition data
nutrition_data, summary, error_message = calculate_meal_health(matched_items)
if error_message:
return "<p>No food items detected</p>", error_message, None, "<p>Analysis not available</p>"
# Generate the food items section
food_items_html = "<ul>"
for item in matched_items:
item_name = item["name"]
if "matched_as" in item:
item_name = f"{item_name} (recognized as {item['matched_as']})"
cals = item["nutrition"]["calories"]
cat = item["nutrition"]["category"].capitalize()
# Choose color based on category
color = "#000000"
if item["nutrition"]["category"] == "healthy":
color = "#007700" # Green
elif item["nutrition"]["category"] == "junk":
color = "#CC0000" # Red
elif item["nutrition"]["category"] == "protein":
color = "#0000CC" # Blue
food_items_html += f'<li style="color:{color};"><strong>{item_name}</strong>: {cals} calories ({cat})</li>'
food_items_html += "</ul>"
# Generate detailed analysis
analysis_img, analysis_html = generate_analysis(matched_items, nutrition_data)
return food_items_html, summary, analysis_img, analysis_html
except Exception as e:
error_msg = f"An error occurred: {str(e)}"
print(error_msg)
return "<p>No food items detected</p>", error_msg, None, "<p>Analysis not available due to an error</p>"
# Create the Gradio interface
def create_gradio_interface():
# Define CSS for the interface
custom_css = """
body {
font-family: 'Arial', sans-serif;
}
h1 {
color: #4a4a4a;
text-align: center;
}
.footer {
text-align: center;
margin-top: 20px;
font-size: 0.8em;
color: #666;
}
.container {
margin: 0 auto;
max-width: 1200px;
}
.tab-content {
padding: 15px;
border: 1px solid #ddd;
border-top: none;
border-radius: 0 0 5px 5px;
}
.nutrition-summary {
background-color: #f9f9f9;
padding: 15px;
border-radius: 5px;
margin-top: 15px;
}
.footer-note {
font-size: 0.9em;
font-style: italic;
margin-top: 30px;
text-align: center;
color: #777;
}
table.table-striped {
width: 100%;
border-collapse: collapse;
}
table.table-striped th, table.table-striped td {
border: 1px solid #ddd;
padding: 8px;
text-align: left;
}
table.table-striped tr:nth-child(even) {
background-color: #f2f2f2;
}
table.table-striped th {
padding-top: 12px;
padding-bottom: 12px;
background-color: #4CAF50;
color: white;
}
"""
# Define theme
theme = gr.themes.Soft(
primary_hue="green",
secondary_hue="blue",
).set(
body_text_color="#333333",
block_title_text_weight="600",
block_border_width="1px",
block_shadow="0px 5px 10px rgba(0, 0, 0, 0.1)",
button_primary_background_fill="#4CAF50",
button_primary_background_fill_hover="#45a049",
)
# Create Gradio blocks
with gr.Blocks(css=custom_css, theme=theme) as demo:
# Header
gr.HTML("""
<div style="text-align: center; max-width: 850px; margin: 0 auto;">
<h1>🧾 Restaurant Bill Nutritional Analyzer 🍔</h1>
<p>Upload a photo of your restaurant bill or receipt, and this tool will analyze what you ate, estimate the nutritional content, and provide a health score.</p>
<p><em>Note: This tool works best with clear images of English-language bills and menus.</em></p>
</div>
""")
# Main content
with gr.Tabs():
# Image upload tab
with gr.TabItem("Upload Receipt Image"):
with gr.Row():
with gr.Column(scale=1):
# Input components
image_input = gr.Image(label="Upload a photo of your restaurant bill")
analyze_btn = gr.Button("Analyze Receipt", variant="primary")
with gr.Column(scale=1):
# Output components
ocr_output = gr.Textbox(label="Extracted Text (OCR)", lines=5)
with gr.Row():
with gr.Column(scale=1):
food_items_output = gr.HTML(label="Detected Food Items")
with gr.Column(scale=1):
nutrition_summary = gr.Textbox(label="Nutrition Summary", lines=4)
with gr.Row():
gr.HTML("<h3>Detailed Nutritional Analysis</h3>")
with gr.Row():
with gr.Column(scale=1):
analysis_chart = gr.Image(label="Analysis Chart")
with gr.Column(scale=1):
analysis_details = gr.HTML(label="Analysis Details")
# Set up the button click event
analyze_btn.click(
fn=process_bill_image,
inputs=[image_input],
outputs=[ocr_output, food_items_output, nutrition_summary, analysis_chart, analysis_details]
)
# Manual text input tab
with gr.TabItem("Enter Food Items Manually"):
with gr.Row():
with gr.Column(scale=1):
text_input = gr.Textbox(
label="Enter what you ate (e.g., 'burger, fries, and a soda')",
lines=3,
placeholder="Example: I had a cheeseburger with fries and a coke for lunch."
)
analyze_text_btn = gr.Button("Analyze Food Items", variant="primary")
with gr.Row():
with gr.Column(scale=1):
text_food_items = gr.HTML(label="Detected Food Items")
with gr.Column(scale=1):
text_summary = gr.Textbox(label="Nutrition Summary", lines=4)
with gr.Row():
with gr.Column(scale=1):
text_analysis_chart = gr.Image(label="Analysis Chart")
with gr.Column(scale=1):
text_analysis_details = gr.HTML(label="Analysis Details")
# Examples for text input
gr.Examples(
examples=[
"I had a burger, fries and a coke",
"For dinner I ordered pizza and ice cream",
"Grilled chicken salad with water",
"Steak, baked potato and broccoli with red wine"
],
inputs=text_input
)
# Set up the button click event
analyze_text_btn.click(
fn=process_text_input,
inputs=[text_input],
outputs=[text_food_items, text_summary, text_analysis_chart, text_analysis_details]
)
# About tab
with gr.TabItem("About"):
gr.HTML("""
<div style="text-align: left; max-width: 850px; margin: 0 auto;">
<h2>About This Tool</h2>
<p>This nutritional analyzer uses OCR (Optical Character Recognition) to extract text from restaurant bills and receipts.
It then uses natural language processing techniques to identify food items and match them to a nutrition database.</p>
<h3>How It Works</h3>
<ol>
<li><strong>Image Processing:</strong> Your uploaded image is enhanced for better text recognition</li>
<li><strong>Text Extraction:</strong> OCR technology reads the text from the image</li>
<li><strong>Food Detection:</strong> NLP algorithms identify food items in the text</li>
<li><strong>Nutrition Matching:</strong> Food items are matched to a nutrition database</li>
<li><strong>Analysis:</strong> Nutritional totals are calculated and a health score is assigned</li>
</ol>
<h3>Limitations</h3>
<p>Please note the following limitations:</p>
<ul>
<li>Works best with clear, well-lit images</li>
<li>Designed primarily for English-language bills and receipts</li>
<li>May not recognize all specialized or regional dishes</li>
<li>Nutritional estimates are approximate and based on standard portions</li>
<li>Health scores are relative indicators and not medical advice</li>
</ul>
<h3>Privacy Notice</h3>
<p>Images uploaded to this tool are processed for the sole purpose of extracting food information.
Images and extracted data are not permanently stored.</p>
<div class="footer-note">
<p>This tool is intended for informational purposes only and is not a substitute for professional nutritional or medical advice.</p>
</div>
</div>
""")
# Footer
gr.HTML("""
<div class="footer">
<p>🍽️ Restaurant Bill Nutritional Analyzer | Built with Gradio and Hugging Face | 2023</p>
</div>
""")
return demo
# Create and launch the app
demo = create_gradio_interface()
# Run the app
if __name__ == "__main__":
demo.launch()