Challenge: Drone Monitoring

Win a mini drone by developing your own automatic animal classifier.

In this challenge you will compete to develop the best automatic animal classifier.

You will be provided with training data (labelled aerial images) and you will have to develop an automatic image classification system. You can leverage existing artificial intelligence/computer vision libraries and frameworks such as Caffe, OpenCV and Theano.

Your task is to train a suitable algorithm to classify aerial images with a high accuracy. At the end of the challenge your algorithm is evaluated based on two metrics, namely, the overall average top 1 error and the overall average top 5 error.

If your algorithm performs the best you will win a mini drone.
Evaluation Criterion: We evaluate the performance of your classification algorithm on a separate ground truth dataset based on the two metrics the overall average top 1 error and the overall average top 5 error.

Evaluation Criteria

Process information
Short introduction
Development of algorithms with support of staff
Winner announcement

Technical information​
The participant needs to bring a laptop and requires basic object oriented programming skills.

Expected result​
Classification algorithm to classify animals.

A small mini drone

Jury:  Anouk Visser & Camiel R. Verschoor.

To Top