Imagenet: A Large-Scale Hierarchical Image Database

 Imagenet: A Large-Scale Hierarchical Image Database

[https://ieeexplore.ieee.org/document/5206848]

2009년 Princeton Univ에서 만든 image database에대한 논문입니다.

[Introduction] 

- Motivation

We believe that a large-scale ontology of images is a critical resource for developing advanced, large-scale content-based image search and image understanding algorithms, as well as for providing critical training and benchmarking data for such algorithms.

- Goal

Our goal is to show that ImageNet can serve as a useful resource for visual recognition applications such as object recognition, image classification and object localization.

[Properties of Imagenet]

1. Scale 

- ImageNet aims to contain in the order of 50 million cleanly labeled full resolution images (500-1000 per synset).

- The current 12 subtrees consist of a total of 3.2 million cleanly annotated images spread over 5,247 categories.

-  To our knowledge this is already the largest clean image dataset available to the vision research community, in terms of the total number of images, number of images per category as well as the number of categories.

 

2. Hierarchy

 

- ImageNet organizes the different classes of images in a densely populated semantic hierarchy.

3. Accuracy

4. Diversity

- have variable appearances, positions, view points, poses as well as background clutter and occlusions.

[Constructing ImageNet]

 

1. Collecting Candidate Images

- ImageNet aims to eventually offer 500-1000 clean images per synset.

- Collect candidate images from the Internet by querying several image search engines.

2. Cleaning Candidate Images

- Used AMT.