Computer science meets biology and medicine to improve health and increase human longevity
by YOuth laboratories

WELCOME TO AGENET

Contests, hackathons and resources for aging biomarker development. Images, blood tests, tissue samples, clinical parameters, and many other data types will be used in regular competitions. Join the AGENET Facebook Group and use #AGENET and #SKINHACK in social networks when participating in AGENET events.

About

AgeNet is an online resource for algorithm developers around the world interested and engaged in aging and healthcare research and developing algorithms for recognition of person's chronological age, biological age, and health status. Like the venerated ImageNet, which helped algorithm developers surpass humans in image recognition using deep learning techniques, AgeNet intends to take biomarker development to the new level.
We will be conducting regular competitions and hackathons on the various data sets starting from facial images to blood tests and genomic data.
Developing accurate predictors of age and health status is important because these exercises will help identify problems at an early stage as well as track the effectiveness of therapeutic interventions and preventive measures.

How it works

Public Databases

Develop an Algorithm for Age Recognition

Test on
Sample Dataset

Beauty.AI
test set
100 pictures

Submit to
AgeNET

Test

Beauty.AI
10000 pictures
(Not Available for training)

Leaderboard

Minimal Requirements

The Neural Network quality should meet the following requirements:
- Age estimation should be done from a single image
- MAE 16 - 22 age <= 4
- MAE 23 - 35 age <= 4.1
- MAE 36 - 45 age <= 6.1
- MAE 46 - 56 age <= 7.8
- MAE 56 - 100 age <= 9.2
- Model weight: <550 Mb

Training Examples