Structured Set Matching Networks for One-Shot Part Labeling

@inproceedings{Choi2018StructuredSM,
  title={Structured Set Matching Networks for One-Shot Part Labeling},
  author={Jonghyun Choi and Jayant Krishnamurthy and Aniruddha Kembhavi and Ali Farhadi},
  booktitle={CVPR},
  year={2018}
}
1. Datasets 1.1. DiPART Dataset 1.1.1 Curation We obtain linedrawing images from image search engines such as Google Image Search and Bing Image Search. We then obtain the list of part names for each object category by asking a set of annotators. We encourage the annotators to list a set of parts that are visually distinctive from one another and unique to that object category, if possible. We then choose the top ten parts. For each image and part, multiple annotators mark a point on the… CONTINUE READING