32Kiran/Real-Estate-Property-Image-Dataset-Lamudi-Facebook-Ohmyhome-PerchanceAI

该数据集旨在训练和评估房地产属性的欺诈过滤系统和图像匹配系统。它包含多样化的房地产图像，包括真实图像和AI生成图像。数据集由三个主要部分组成：1. AI生成图像：使用Perchance AI生成，涵盖11种属性类型（如公寓、别墅）、10种建筑风格、2种属性类别（内部/外部）和多个特征（如卧室、立面），总计2000张图像（1000张训练，1000张测试）。2. 真实图像：从Facebook Marketplace爬取，包含1000个属性和8158张图像，数据格式为CSV，包括图像路径、标题、价格等信息。3. 版权图像：从Lamudi和Ohmyhome爬取，包括无EXIF数据的图像（3000张）和水印图像（1500张，分为透明logo、文本水印和logo水印）。数据集考虑了伦理问题，如数据隐私、版权合规和数据偏见，为房地产图像匹配和欺诈检测提供了有价值的资源。

This dataset is designed to train and evaluate a fraud filter system and image matching system for real estate properties. It comprises a diverse range of real estate images, including authentic images and AI-generated images. The dataset consists of three main components: 1. AI-Generated Images: Generated using Perchance AI, covering 11 property types (e.g., Condominium, House and Lot), 10 architectural styles, 2 property categories (Interior/Exterior), and multiple features (e.g., bedroom, facade), totaling 2000 images (1000 training, 1000 testing). 2. Authentic Images: Scraped from Facebook Marketplace, containing 1000 properties and 8158 images, with data in CSV format including image paths, title, price, etc. 3. Copyrighted Images: Scraped from Lamudi and Ohmyhome, including images without EXIF data (3000 images) and watermarked images (1500 images, divided into transparent logo, text watermark, and logo watermark). Ethical considerations such as data privacy, copyright compliance, and data bias are addressed, making this dataset a valuable resource for real estate image matching and fraud detection.