Business Problem

• Property Finder's growing database of property listings needs help maintaining image quality and accuracy.
• High-quality visuals are crucial in real estate to attract and retain customers.
• Correct images can erode customer trust, harm user experience, and damage the platform's credibility.
• Consistent or low-quality photos can lead to satisfaction among potential buyers or renters, resulting in lost business opportunities for Property Finder and its real estate partners.
• Manual image review processes could be more efficient and efficient.
• An automated solution is needed to ensure image quality and consistency.
• Maintaining high-quality visuals is critical to delivering a trustworthy and professional platform for users.

Solution

Image Quality Validation API: Validates image quality using AI models built on AWS Bedrock, ensuring that images meet the required clarity, resolution, and quality standards FastAPI for handling requests, AWS Lambda for serverless execution, AWS Bedrock for AI-driven image validation.

Image Enhancement API: This API enhances images using AI-based techniques, including upscaling and improving clarity for better visual appeal. It uses FastAPI for request management, AWS Lambda for processing, and AWS Bedrock for applying AI techniques to enhance images.

Image Detail Extraction API: Using machine learning models on AWS Bedrock, extract key metadata from property images, such as room details or views. FastAPI is used for efficient API processing, AWS Lambda executes tasks serverlessly, and AWS Bedrock is used for detail extraction through AI.

Text and Image Data Comparison API: This API compares extracted image details with the textual property listing, ensuring data consistency by providing a match/mismatch percentage. FastAPI directs requests, AWS Lambda processes, and the LLM Model (To Be Discovered) performs language model-based comparison.

Architecture

• AWS S3 (Simple Storage Service): Stores property images uploaded by users for further processing.
• Users: Users interact with the system through APIs to upload images and trigger processing.
• FastAPI: Handles incoming API requests for image validation, extraction, and enhancement.
• Git: Version control for source code, enabling seamless deployment and updates using Docker.
• AWS Lambda: Serverless execution environment for handling API requests and processing tasks.
• Bedrock Claude-V3.5: AI-driven service for image validation, feature extraction, and text-image comparison.
• Project Engineering: Represents the logical workflow between Lambda functions and AI models for processing.
• Ray Serve / ComfyUI on EC2: AI-powered image enhancement deployed on EC2 for advanced image upscaling and quality improvement
• RDS (Relational Database Service): Stores intermediate data and results from image enhancement tasks.
• Monitoring and Logging (AWS CloudWatch): This function logs, monitors, and triggers alerts to ensure system health and performance.
• User Management: Tracks and manages user activity and access within the system.