app_seg_llm.py

import streamlit as st
from PIL import Image 
import os
import anthropic
import base64
import numpy as np
from dotenv import load_dotenv
import cv2
import tempfile
import easyocr 
import pytesseract
load_dotenv() 


from yolo_functions import segment_large_image_with_tiles , usable_data , plot_differences_on_image1 , system_prompt_4 , blueprint_analyzer
from ultralytics import YOLO 
from openai import OpenAI
import os



client = anthropic.Anthropic(
    # api_key="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",
    api_key = os.getenv('ANTHROPIC_API_KEY')
)

openai_client = OpenAI(api_key=os.getenv("OPENAI_API_KEY")) 


def encode_image(image_path):
    # with open(image_path, "rb") as image_file:
        # return base64.b64encode(image_file.read()).decode("utf-8")
    return base64.b64encode(image_path.getvalue()).decode("utf-8")

def chat_claude(prompt , image1 , image2 ) :
    # print("image 1"  , image1)
    image1_data = encode_image(image1) 
    # print("image 1 data" , image1_data)
    image2_data = encode_image(image2) 
    message = client.messages.create(
        model="claude-3-opus-20240229",
        max_tokens = 4096,
        temperature=0,
        messages=[
            {
                "role": "user",
                "content": [
                    {
                        "type": "text",
                        "text": "Image 1:"
                    },
                    {
                        "type": "image",
                        "source": {
                            "type": "base64",
                            "media_type": "image/jpeg",
                            "data": image1_data,
                        },
                    },
                    {
                        "type": "text",
                        "text": "Image 2:"
                    },
                    {
                        "type": "image",
                        "source": {
                            "type": "base64",
                            "media_type": "image/jpeg",
                            "data": image2_data,
                        },
                    },
                    {
                        "type": "text",
                        "text": f"{prompt}"
                    }
                ],
            }
        ],
    )
    return message.content[0].text


prompt = """Given 2 construction blueprints your task is to analyze carefully both blueprints and point out difference for following categories -
1. Strcutural grid.
2. Layout Areas - rooms , balcony , porch , staircase , elevator etc.
3. Interior changes or optimization.
Summarize all the difference in paragraph concisely.
"""
st.set_page_config(layout = "wide")
uploaded_files = st.file_uploader("Upload 2 image to compare", accept_multiple_files=True)
# import pdb; pdb.set_trace()
# print("upladed file length" , len(uploaded_files))
if len(uploaded_files) !=0  :
    temp_dir = tempfile.TemporaryDirectory() 
    # print(temp_dir)
    i = 0
    for one_file in uploaded_files :
        if i == 0 :
            img1 = Image.open(one_file)  
            sv_path_1 = temp_dir.name + "/img1.jpg"  
            img1.save(sv_path_1) 
            # print("uploaded file" , one_file)
            # print("img1" , img1)

            tmp_img1 = one_file  
            # print("tmp_img1" , tmp_img1)  
            st.image(img1)
            i = i + 1
        if i == 1 :
            img2 = Image.open(one_file) 
            sv_path_2 = temp_dir.name + "/img2.jpg"  
            img2.save(sv_path_2) 
            tmp_img2 = one_file     
            st.image(img2)
            i = i + 1

    col1 , col2 = st.columns(2)
    col1.header("LLM")
    col2.header("Seg-LLM !")
    # import pdb; pdb.set_trace()
    llm_ans = chat_claude(prompt , tmp_img1 , tmp_img2)
    print(llm_ans)
    col1.write(llm_ans)
    #################### yolo segment from here ################
    model = YOLO("best.pt") 

    final_output_1, class_mask_dict_1 = segment_large_image_with_tiles(
    model,
    # large_image_path=img_1_path,
    # large_image_path= tmp_img1 ,
    large_image_path= sv_path_1 ,
    tile_size=1080,
    overlap=120,
    alpha=0.4,
    display=True
    )
    final_output_2, class_mask_dict_2= segment_large_image_with_tiles(
    model,
    large_image_path=sv_path_2,
    tile_size=1080,
    alpha=0.4,
    display=True
    )
    label_dict = {0: 'EMP', 1: 'balcony_area', 2: 'bathroom', 3: 'brick_wall', 4: 'concrete_wall', 5: 'corridor', 6: 'dining_area', 7: 'door', 8: 'double_window', 9: 'dressing_room', 10: 'elevator', 11: 'elevator_hall', 12: 'emergency_exit', 13: 'empty_area', 14: 'lobby', 15: 'pantry', 16: 'porch', 17: 'primary_insulation', 18: 'rooms', 19: 'single_window', 20: 'stairs', 21: 'thin_wall'}
    img1_results = {}
    for key in class_mask_dict_1.keys():
        img1_results[label_dict[key]] = class_mask_dict_1[key]
    img2_results = {}
    for key in class_mask_dict_2.keys():
        img2_results[label_dict[key]] = class_mask_dict_2[key]
    image_1 , image_2 = img1 , img2 
    width, height = image_1.width, image_1.height 
    image_1_data = usable_data(img1_results, image_1) 
    image_2_data = usable_data(img2_results, image_2) 

    lines_1, text_data_1 = blueprint_analyzer(sv_path_1)
    lines_2, text_data_2 = blueprint_analyzer(sv_path_2)

    user_prompt_3 = f"""I have two construction blueprint images, Image 1 and Image 2, and here are their segmentation results (with bounding boxes, centers, and areas). Please compare them and provide a short Markdown summary of the differences, ignoring any objects that match in both images:

        Image 1:
        image: {image_1}
        segmentation results: {image_1_data}
        grid lines: {lines_1}
        ocr results: {text_data_1}
        Image 2:
        image: {image_2}
        segmentation results: {image_2_data}
        grid lines: {lines_2}
        ocr results: {text_data_2}

        Please:
        Compare the two images only in terms of differences—ignore any objects that match (same label and near-identical center).
        For objects missing in Image 2 (but present in Image 1), or newly added in Image 2, indicate their relative position using known areas or approximate directions. For instance, mention if the missing doors were “towards the north side, near the elevator,” or if new walls appeared “in the southeastern corner, near the balcony.”
        Summarize any changes in labels or text, again without giving raw bounding box or polygon coordinate data.
        Provide your final output in a short, clear Markdown summary that describes where objects have changed.
        Mention if there are text/label changes (e.g., from an OCR perspective) in any particular area or region
    """



    

    completion = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[
            {"role": "system", "content": system_prompt_4},
            {
                "role": "user",
                "content": user_prompt_3
            }

        ]
    )

    print(completion.choices[0].message.content)