Overview

How can we judge if a pair of objects are stylistically compatible with each other? We investigated how Siamese networks can be used efficiently for assessing the style compatibility between images of furniture items.
To help us in this task, we collect and present a dataset of 90,298 images from 6 categories of furniture across 17 different styles. The images have corresponding text metadata containing information as manufacturer, size, weight, materials, subcategories (e.g folding chair, dining chair, armchair) and, most importantly, the style. More details can be found in our paper.

Statistics

Distribution of furniture types in the dataset.

In our dataset, there are 6 different furniture types. The furniture in the dataset is categorized into beds, chairs, dressers, lamps, sofas and tables.

Style distribution for the entire dataset

Our dataset has a large variety of furniture styles: Asian, Craftsman, Industrial, Modern, Southwestern, Tropical, Beach, Eclectic, Mediterranean, Rustic, Traditional, Victorian, Contemporary, Farmhouse, Midcentury, Scandinavian, Transitional

Paper

                    
                      @inproceedings{aggarwal2018learning,
                        title={Learning Style Compatibility for Furniture},
                        author={Aggarwal, Divyansh and Valiyev, Elchin and Sener, Fadime and Yao, Angela},
                        booktitle={German Conference on Pattern Recognition},
                        pages={552--566},
                        year={2018},
                        organization={Springer}
                      }