Presentation Title: Amenity Detection &  Inventory Tracking using Deep Learning techniques

Presenter(s): Faiza Ayoun, Nilisha Makam Prashantha, & Sangamithra Murugesan

Abstract: Short-term rentals, facilitated by platforms like Airbnb and Vrbo, are lucrative for small property owners, providing financial benefits and affordable lodging. However, homeowners express concerns about property conditions post-checkout. This research develops machine-learning tools to address the challenges of Airbnb’s 6.6 million listings’ amenity management. A web application was created that uses deep-learning and computer vision techniques to automate amenity identification and cataloging, assist hosts in tracking, and present relevant information, ultimately improving the property management experience.

This research, guided by the CRISP-DM methodology, involved collecting data using open-source image datasets and preparing them for modeling. Key stages included cleaning, image annotation, and addressing class imbalances. Object detection models like Detectron2-Mask R-CNN, YOLOv7, EfficientDet, RetinaNet, and DETR were evaluated, with the best model deployed in a Python-based web app built using Streamlit and Google-Cloud SQL server.

‘Amenitrack,’ the web application, predicts amenities and their count based on uploaded images, providing a platform for amenity analysis through charts and maps. It offers a comprehensive solution for platforms like AirBnB, enhancing utilities and aiding hosts in managing amenities at scale.

The application’s analytics assists hosts in monitoring missing inventory and maintaining reserve inventory. Guests benefit from transparent information sharing and simplified amenities-based property searches. Beyond hospitality, this application can be adaptable to inventory management in other domains. Apart from providing the design and codebase of the application, the team also presented all the artifacts about the progress of the application development, such as reports, presentations, and demo videos.”

Link to Recorded Presentation: https://youtu.be/j3hdN1xQNBE

 

9 thoughts on “(2024) Amenity Detection & Inventory Tracking using Deep Learning techniques

  1. Anonymous

    This application looks promising with the integration of state-of-the-art models and can help lot of rental platforms and small business to keep track of the inventory.

    • Anonymous

      Thank you. We’re glad you found our project with potential.

  2. Zoulikha

    I have friends who own Airbnb properties and I have seen their frustration when they find something of value broken missing and cannot identify when it happened. This work has the potential to solve this issue, which will encourage other owners to put their property out for rent, potentially releasing the increasing need of short term rentals. Great work!

    • Anonymous

      Thank you, your encouragement means a lot to us, and we will try to continue refining our solution to meet the needs of property owners like your friends.

  3. Shilpana Sathyanarayana

    Interesting work and well presented. Very methodically done.

    • Anonymous

      We appreciate your comment.

  4. Ibtissem

    A great solution to a real problem. Would love to see it materialized! I can see how this solution would solve many issues on the ground. Great work!

    • Faiza Ayoun

      Thank you! We appreciate your positive input.

  5. Shruthi Sathish

    Great job!! Your innovative approach and thorough exploration of the application of deep learning models were impressive. Keep up the excellent work in advancing AI and computer vision!

Leave a Reply

Your email address will not be published. Required fields are marked *