Code Preview:
import re
import openpyxl
import cv2 # Import OpenCV for real-time capture
from PIL import Image
import torch
from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor
import numpy as np
from google.colab import files
# Load the Qwen2 model and processor
model = Qwen2VLForConditionalGeneration.from_pretrained(
"Qwen/Qwen2-VL-2B-Instruct",
torch_dtype="auto",
device_map="auto",
)
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
# Create a new Excel workbook
workbook = openpyxl.Workbook()
sheet = workbook.active
sheet.title = "Product Analysis"
# Add headers
headers = ["Product Name", "Category", "Quantity", "Count", "Expiry Date", "Freshness Index", "Shelf Life"]
sheet.append(headers)
# Regular expression patterns to extract data
packaged_product_pattern = r"Product Name: (.*)\n - Product Category: (.*)\n - Product Quantity: (.*)\n - Product Count: (.*)\n - Expiry Date: (.*)"
fruits_vegetables_pattern = r"Type of fruit/vegetable: (.*)\n - Freshness Index: (.*)\n - Estimated Shelf Life: (.*)"
# Let user upload the image
uploaded = files.upload()
# Iterate over all uploaded files
for image_name in uploaded.keys():
# Open the image using PIL
image = Image.open(image_name)
image = image.resize((512, 512))
# Prepare the text prompt for predicting product details and freshness
messages = [
{
"role": "user",
"content": [
{
"type": "image",
"image_url": "Captured from webcam" # Description for internal use
},
{
"type": "text",
"text": """This image contains fruits, vegetables, or packaged products.
Please analyze the image and provide:
- For packaged products:
- Product Name
- Product Category
- Product Quantity
- Product Count
- Expiry Date (if available)
- For fruits and vegetables:
- Type of fruit/vegetable
- Freshness Index (based on visual cues)
- Estimated Shelf Life"""
}
]
}
]
# Prepare the text prompt for processing
text_prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
# Process the image and prompt for model input
inputs = processor(
text=[text_prompt],
images=[image],
padding=True,
return_tensors="pt"
)
# Move inputs to GPU if available
inputs = inputs.to("cuda" if torch.cuda.is_available() else "cpu")
# Generate output from the model
output_ids = model.generate(**inputs, max_new_tokens=1024)
# Decode the generated output
generated_ids = [
output_ids[len(input_ids):]
for input_ids, output_ids in zip(inputs.input_ids, output_ids)
]
output_text = processor.batch_decode(
generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True
)[0]
# Print the output text for the current image
print(f"Analysis Output for the captured frame:")
print(output_text)
# Extract packaged product information
packaged_product_match = re.search(packaged_product_pattern, output_text)
fruits_vegetables_match = re.search(fruits_vegetables_pattern, output_text)
if packaged_product_match:
product_name = packaged_product_match.group(1).strip()
category = packaged_product_match.group(2).strip()
quantity = packaged_product_match.group(3).strip()
count = packaged_product_match.group(4).strip()
expiry_date = packaged_product_match.group(5).strip()
else:
product_name = category = quantity = count = expiry_date = "N/A"
if fruits_vegetables_match:
# Overwrite product fields with fruit/vegetable data if applicable
product_name = fruits_vegetables_match.group(1).strip()
category = "Fruit/Vegetable"
freshness_index = fruits_vegetables_match.group(2).strip()
shelf_life = fruits_vegetables_match.group(3).strip()
else:
freshness_index = shelf_life = "N/A"
# Insert row into Excel
sheet.append([product_name, category, quantity, count, expiry_date, freshness_index, shelf_life])
# Save the workbook
workbook.save("product_analysis.xlsx")
print("Data saved to product_analysis.xlsx")
