Insights About Places with Deep Learning Computer Vision • Chanuki Illushka Seresinhe • YOW! 2023

"Explore the application of deep learning computer vision in property valuations and analytics, from metadata labeling to transfer learning and beyond, to provide more accurate insights and better understanding of consumer preferences."

Key takeaways
  • Use of computer vision in property valuations and analytics
  • Importance of metadata and labeling in deep learning models
  • Use of transfer learning for improved model accuracy
  • Development of a community interest company to monetize property data
  • Integration of convolutional neural networks (CNNs) with other AI models
  • Importance of considering climate risks and flood risks in property valuations
  • Use of natural language processing (NLP) to analyze property descriptions and sentiment analysis
  • Development of a walking app for outdoor exploration using property data
  • Use of active learning to improve model performance and reduce labeling costs
  • Importance of understanding consumer preferences and beautiful places in property valuations
  • Use of Jay Appleton’s prospect and refuge theory to analyze the impact of urban design on well-being
  • Development of a large-scale, high-resolution scene recognition model for property valuations
  • Use of machine learning to create 3D models and virtual tours of properties
  • Importance of integrating technology with human judgment in property valuations