### Hemanth Venkateswara

I am a researcher in machine learning and computer vision. I work on deep learning based domain adaptation. I enjoy Sanskrit poetry.

# Office-Home Dataset

The Office-Home dataset has been created to evaluate domain adaptation algorithms for object recognition using deep learning. It consists of images from 4 different domains: Artistic images, Clip Art, Product images and Real-World images. For each domain, the dataset contains images of 65 object categories found typically in Office and Home settings.

Figure 1: Sample images from the Office-Home dataset. The dataset consists of images of everyday objects organized into 4 domains; Art: paintings, sketches and/or artistic depictions, Clipart: clipart images, Product: images without background and Real-World: regular images captured with a camera. The figure displays examples from 16 of the 65 categories.

## Data Collection

The images in the dataset were collected using a python web-crawler that crawled through several search engines and online image directories. This initial run searched for around 120 different objects and produced over 100,000 images across the different categories and domains. These images were then filtered to ensure that the desired object was in the picture. Categories were also filtered to make sure that each category had at least a certain number of images. The latest version of the dataset contains around 15,500 images from 65 different categories.

Domain Min: # Min: Size Max: Size Acc.
Art 15 117 $$\times$$ 85 pix. 4384 $$\times$$ 2686 pix. 44.99 $$\pm$$ 1.85
Clipart 39 18 $$\times$$ 18 pix. 2400 $$\times$$ 2400 pix. 53.95 $$\pm$$ 1.45
Product 38 75 $$\times$$ 63 pix. 2560 $$\times$$ 2560 pix. 66.41 $$\pm$$ 1.18
Real-World 23 88 $$\times$$ 80 pix. 6500 $$\times$$ 4900 pix. 59.70 $$\pm$$ 1.04

Statistics for the Office-Home dataset. Min: # is the minimum number of images of each object for the specified domain. Min: Size and Max: Size are the minimum and maximum image sizes in the domain. Acc: is the classification accuracy using a linear svm (LIBLINEAR) classifier with 5-fold cross-validation using deep features extracted from the VGG-F deep network.

## Object Categories

The 65 object categories in the dataset are:

Alarm Clock, Backpack, Batteries, Bed, Bike, Bottle, Bucket, Calculator, Calendar, Candles,
Chair, Clipboards, Computer, Couch, Curtains, Desk Lamp, Drill, Eraser, Exit Sign, Fan,
File Cabinet, Flipflops, Flowers, Folder, Fork, Glasses, Hammer, Helmet, Kettle, Keyboard,
Knives, Lamp Shade, Laptop, Marker, Monitor, Mop, Mouse, Mug, Notebook, Oven, Pan,
Paper Clip, Pen, Pencil, Postit Notes, Printer, Push Pin, Radio, Refrigerator, ruler,
Scissors, Screwdriver, Shelf, Sink, Sneakers, Soda, Speaker, Spoon, Table, Telephone,
Toothbrush, Toys, Trash Can, TV, Webcam


The dataset was curated by Jose Eusebio. Please use the below citation to refer to the dataset.

## Bibtex

@inproceedings{venkateswara2017Deep,
title={Deep Hashing Network for Unsupervised Domain Adaptation},
author={Venkateswara, Hemanth and Eusebio, Jose and Chakraborty, Shayok and Panchanathan, Sethuraman},
booktitle={({IEEE}) Conference on Computer Vision and Pattern Recognition ({CVPR})},
year={2017}
}