Families In the Wild: A Kinship Recognition Benchmark

• 2019-10-11 4th Recognizing Families In the Wild Challenge (Web, CV News,, ACM Press Release)
• 2019-05-20 FIW to organize premiere Kaggle Competition [kaggle]
• 2019-01-21 Tutorial on kinship recognition at 2019 CVPR
  
• 2019-01-16 Tutorial on kinship recognition at 2019 FG
  
• 2018-10-27 Presented tutorial on kinship recognition at 2018 ACM MM [slides]
• 2018-08-09 3rd Recognizing Families In the Wild (2018 RFIW) at Automatic Face & Gesture (2018 FG)
• 2018-10-01 Sample code and tools updated on github [code]
• 2018-05-15 Kinship Classification through Latent Adaptive Subspace presented at 2018 FG in Xi’an, China [paper]
• 2018-05-14 2018 Workshop on RFIW (version 2) at Automatic Face & Gesture (2018 FG)
• 2018-03-14 Visual Recognition of Families In the Wild (FIW) published in TPAMI [paper]
• 2017-10-27 2017 RFIW Data Challenge Workshop at 2017 ACM MM in Mountain View, CA [proceedings, paper]
• 2017-05-31 Verification FIW with Marginalized Denoising Metric Learning presented at 2017 FG [poster, paper]
• 2017-05-30 Poster presented at 2017 FG in Washington DC [poster]
• 2017-03-30 Built Project Page (this)
• 2016-12-11 Poster presented at 2017 FG in Washington DC [poster]
• 2016-05-30 Presented at 2016 New England Computer Vision Workshop at BU [extended abstract, slides]
• 2016-10-15 FIW presented at ACM MM 2016 in Amsterdam, Netherlands [paper, poster]

Families In the Wild (FIW) is the largest and most comprehensive image database for automatic kinship recognition. Our motivation is to provide the resource needed for kinship recognition technologies to transition from research-to-reality. with over 11,932 family photos of 1,000 families FIW closely reflects the true data distribution of families worldwide (see Database for more information). There are many directions for FIW to take throughout the machine vision and related research communities (e.g., in relation to benchmarks (see Challenges and Results for details), new benchmarks, generative models, multi-modal learning, to name a few). In terms of its practical value, many could benefit from FIW as well, such as the consumer (e.g. automatic photo library management), scholar (e.g. historic lineage & genealogical studies), analyzer (e.g. social-media-based analysis), investigator (e.g. missing persons and human traffickers).

Last May I unfortunately had to present a section of our tutorial remotely (i.e., always much better to be at the conference, especially for a talk). There is an up-side to this, however– while in the midst of giving my laptop a New Year cleanup, I stumbled upon the recording I sent in case we had issues connecting for my part of the talk. Considering most appreciate a speaker to be present, and never mind at least connected remotely for questions, this was a last resort that we narrowed down to just cover some part of the FIW data and benchmarks. Sure enough, perhaps some might find it useful. Any and all feedback is more than welcome (in fact, would be much appreciated). It is important to note the date of the video (i.e., statistics mentioned were from a previous version of FIW). Also, best to reference PAMI and other papers for up-to-date benchmarks. Nonetheless, besides a few tables, the gist remains as is. Regardless, I hope you enjoy 🙂

Click frame above to start video. If unavailable, here is the direct link: https://1drv.ms/v/s!AkDk_XdBkJ9whucKtLWTfreWddVWkw?e=toqkOF

Below we list other kinship recognition datasets. A more detailed comparison of the datasets can be found in the paper.

• KinFaceW: 2,000 face pairs of 4 different relationship types (i.e., parent-child)
• TSKinFace: Tri-subject pairs (i.e., mother+father-child)
• Family101: Image collection with 101 families (i.e., tree-like labels)
• UB Kin dataset: Image collection of face pairs for kinship recognition.

Citation

@inproceedings{fiwpamiSI2018,

Author = {Robinson, J P and Shao, M and Wu, Y and Liu, H and Gillis, T and Fu, Y},

Booktitle = {IEEE Transactions on pattern analysis and machine intelligence},

Title = {Visual Kinship Recognition of Families in the Wild},

        Keywords={Large-Scale Image Dataset, Kinship Verification, Family Classification, Semi-Supervised Clustering, Deep Learning},

Year = {2018}}

For questions and result submission, please contact Joseph Robinson at robinson.jo@husky.neu.edu