# Automatic and Accurate Classification of Hotel Bathrooms from Images with Deep Learning

**Authors:** Hakan Temiz  
**arXiv:** [2306.07727](https://arxiv.org/abs/2306.07727)  
**Date:** 2023-06-13  
**Summary page:** https://www.summarizepaper.com/en/paper/2306.07727v1/

## Key points

- Hotel guests' satisfaction is heavily influenced by the condition of bathrooms
- Guests often rate hotels and post images related to their positive or negative experiences, with a focus on the bathroom
- Comments and visuals shared by guests can affect prospective guests' decisions when choosing a hotel
- A study was conducted to develop deep learning algorithms that accurately classify hotel bathrooms based on images as satisfactory or unsatisfactory
- The models were trained using four different image sizes: 128, 256, 512, and 1024 pixels in both directions
- The top-performing algorithm achieved an accuracy rate of 92.4% and an AUC score of 0.967 in classifying bathroom images as satisfactory or unsatisfactory
- This study provides valuable insights for hotel owners who want to improve customer satisfaction by addressing issues related to their bathrooms
- Utilizing these deep learning algorithms can help identify areas that need improvement before receiving negative feedback from customers
- This study contributes to the development of computer vision technology for image classification tasks beyond traditional applications such as object recognition or facial recognition systems.

## Summary

The satisfaction of hotel guests is heavily influenced by the condition of the bathrooms, which are often the subject of complaints. To share their experiences, guests rate hotels and post images that showcase their positive or negative experiences. These images are often related to the bathroom, as guests tend to use them to express their satisfaction or dissatisfaction with this particular area. The comments and visuals shared by guests can potentially affect prospective guests' decisions when choosing a hotel. To address this issue, a study was conducted to develop deep learning algorithms that can accurately classify hotel bathrooms based on images as satisfactory (good) or unsatisfactory (bad). The models were trained using four different image sizes: 128, 256, 512, and 1024 pixels in both directions. The results showed that both algorithms performed exceptionally well in classifying bathroom images with high accuracy rates. The top-performing algorithm achieved an accuracy rate of 92.4% and an AUC score of 0.967. Other metrics also proved the success of these algorithms in accurately identifying satisfactory and unsatisfactory bathrooms. This study provides valuable insights for hotel owners who want to improve customer satisfaction by addressing issues related to their bathrooms. By utilizing these deep learning algorithms, they can quickly identify areas that need improvement and take corrective measures before receiving negative feedback from customers. Additionally, this study contributes to the development of computer vision technology for image classification tasks beyond traditional applications such as object recognition or facial recognition systems.