HomeWhy Challenges?All ChallengesCreate Your Own ChallengeContributorsForum
Sign in / Register

Downloading the data

We provide 3 methods for acquiring the dataset used in this challenge:

  1. tEPIS platform: an online slide-sharing solution by Philips through which the participants can access the full data (discontinued)
  2. Downloading the entire dataset using drive.google.com
  3. CVEDIA platform: a cloud-based platform for public access to the CAMELYON16 dataset
  4. Downloading the entire dataset using academictorrents.com


1) tEPIS platform

tEPIS is a web-based tool suitable for Pathology image management and sharing through which challenge participants can access the data. The main advantage of this platform is that participants cannot only access the full data but also query the system for individual tiles from differing resolution levels, circumventing the need for storing the data locally. Additionally, this web platform contains a visualization component so participants can view the slides through their web browser. Within 24 hours after registration, participants will receive a username and password for accessing tEPIS.
We provide a Matlab and Java client for the tEPIS. You can access these clients together with their documentation from this link.


2) Google drive

Participants can also download the entire dataset on their local machine using this link.
The folder containing the dataset is only shared privately with the participants. Please request access to the dataset preferably using a Gmail account. We recommend you to open a free google drive account first and add the folder to your own account. This way downloading can be done much easier in batch. Please note that the items shared with you in Google drive do not count toward your storage limit.


CVEDIA is a cloud-based platform that allows you export the challenge data for use in your Machine Learning / Deep Learning framework. You can select images from multiple datasets, apply selective augmentations and filters, and download them directly onto your server. Basically, you can generate thousands of totally random image crops from one slide.
You may access CAMELYON16 data using this link.
We recommend the participants to watch the tutorial provided by CVEDIA showing the features of the platform and instructions for exporting the data.

4) Academictorrents

This service is designed to facilitate storage of all the data used in research, including datasets as well as publications. The data has been divided into several parts:

Training dataset for normal images:

Training dataset for tumor containing images:

Ground Truth data:

Test dataset:

Camelyon16 Challenge Google Group

We have created a private Google group for our challenge. This group is created to enable discussion about different topics relevant to the challenge. It is also the best way to hear about our announcements. This forum is directly available on Camelyon16 grand-challenge site under tab Discussion Group. In order to register in Camelyon16 Google group, the participants need to have a Google account. Therefore, we would like to ask the participants to provide us with an additional email address in Gmail (Not required for those who registered using a Gmail address).



Consortium for Open Medical Image Computing © 2012-