Our ambition


In January 2018, Naturalis Biodiversity Center, Observation International and COSMONiO started a long-term partnership with the goal to advance targeted biodiversity work areas through the use of Artificial Intelligence. The AI Nature initiative has been set up as a means to realize the mission.

The targeted work areas are:

  • Biodiversity and nature research on both a professional and an amateur level
  • Management of natural history collections
  • Nature management and nature conservation
  • Nature education


The partners believe effective advances in the biodiversity work areas can be best achieved by:

  • using a scientific method and quality, freely accessible deep-learning software for the development and innovation of recognition architectures (types of recognition models);
  • precise insight into the nature of knowledge-related problems and opportunities within targeted biodiversity work areas;
  • the co-creation with work area partners of generally applicable solutions. Recognition models form the heart of these solutions. User software ensures access to and proper usage of these models;
  • following two approaches, i.e. a Do-it-ourselves approach and an Enable-other-parties approach.


The partners' ambition is:

  • to help interested organizations within the defined biodiversity working areas to optimize their operational processes with standardized and modularized solutions that encompass (bespoke) recognition models and easy-to-use user software;
  • to share knowledge about the advantages and disadvantages of using AI technology within operational processes;
  • to carefully extend the consortium with technology and data partners.

Do it ourselves approach

On this path, the partnership develops and implements advancements in close cooperation with partners within targeted biodiversity work areas. We share the knowledge from these trajectories, implement solutions for a fair price, and make more progress by actively working on partnership growth.

This site features the initial results from the Do-it-ourselves path. It showcases the first advancements which have been realized or are in the prototype phase; the solution building blocks that underlie these advancements, and, the research that has been done on recognition architectures and the implemented deep-learning facility.

Enabling other parties approach

The Enable-other-parties path encourages and facilitates social, scientific, and software development parties to contribute to the stated mission by:

  • providing tools for solution development i.e.:
  1. open access to recognition models through APIs
  2. open access to key components of the deep-learning facility for the biodiversity research community and allowing them to participate in developing recognition architectures.
  • encouraging and enabling third parties to share data, ideas, knowledge, and solutions