matkowski-voy / The-Nipple-Areola-Complex-for-Criminal-Identification

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The-Nipple-Areola-Complex-for-Criminal-Identification

Paper

Wojciech Michal Matkowski, Krzysztof Matkowski, Adams Wai Kin Kong and Cory Lloyd Hall. The Nipple-Areola Complex for Criminal Identification. In International Conference on Biometrics (ICB), June 2019, DOI: 10.1109/ICB45273.2019.8987341.

Paper is on arXiv here
Paper is on IEEE Xplore here

Dataset

NTU-Nipple-v1 dataset contains 2732 images of 428 different male nipple-areolae (NAC) from 2 sessions in relatively well-controlled environment.

How to acquire dataset?

To acquire the NTU-Nipple-v1 dataset (which was used in the paper), download and fill in the "Data Release Agreement.pdf" file. Print the agreement and sign on page 2. Scan the signed copy and send back to matk0001@e.ntu.edu.sg, xpxu@ntu.edu.sg or adamskong@ntu.edu.sg with title "Application for NTU-Nipple-v1 dataset". A download link to the corresponding dataset will be send to you once after we receive the signed agreement file.

Image examples

Examples of images from NTU-Nipple-v1 dataset. Note that these example NAC images are resized to the same size. In NTU-Nipple-v1 dataset, NAC images have different (but still square) sizes.
alt text
First row: session 1
Second row: session 2
NOTE: in this figure, images are resized down to the same size. In NTU-Nipple-v1 image sizes vary and the median size is 179 by 179 pixels.

Results

Currently the highest identification accuracies (under the protocol described in the paper) are:
75.75% at rank-1
94.46% at rank-15
96.54% at rank-30

alt text

Questions

If you have any questions about the paper please email me on matk0001@e.ntu.edu.sg or maskotky@gmail.com or wojciech.matkowski@ntu.edu.sg