Our tree data has been updated again. Four major improvements were made in this update: a national update based on the 2025 summer aerial photos, an extension of the detection model with greenhouse recognition, the addition of monumental trees to the dataset and better detection of undergrowth.
Update with summer aerial photos 2025
The entire tree database was reprocessed based on the most recent 2025 summer aerial photographs. These aerial photographs are collected annually during the period from approximately May to September.
This update makes the tree data even more closely match the current situation in the field. Trees that have since disappeared are automatically marked in the dataset and new situations become more visible in the analysis.
Improved detection: greenhouses excluded from tree detection
A second improvement is in the detection model that automatically identifies trees.
We have developed a proprietary model that recognises greenhouses (greenhouses). This recognition is now integrated into the tree detection process. As a result, greenhouses are automatically excluded from tree detection.
This is relevant because greenhouses in aerial photo and height analyses sometimes have properties similar to trees. They have height and are often transparent, allowing vegetation within the greenhouse to be visible in aerial photographs. As a result, in some cases they could be more likely to be interpreted as trees.
By explicitly including greenhouse detection in the model, these objects are now automatically filtered. This ensures a more consistent tree file.
Monumental trees added
We also added a new attribute layer: monumental trees.
This information is linked to our tree data and is also made available through the API. This allows users to easily see if a tree is registered as monumental.
The monumental trees are based on open data from De Bomenstichting. This organisation records and manages information on monumental trees in the Netherlands.
A tree can be classified as monumental if it has special value, for example in terms of:
- cultural history
- landscape and heritage
- ecology
- age
- special appearance or type
Linking this dataset spatially to our tree inventory creates an additional layer of information that can be valuable to policymakers, researchers and public space managers.
Optimised algorithm for better undergrowth detection
In addition to the above improvements, the algorithm has been further optimised, with the specific aim of better detecting undergrowth. Undergrowth means lower vegetation located under or between tree crowns, such as shrubs, bushes and smaller storage. This type of vegetation was more difficult to distinguish from the surrounding environment in earlier versions of the model. With modifications to the model architecture and refinement of the detection parameters, it is now possible to more accurately identify and classify this undergrowth. This provides a more detailed and complete picture of the overall green structure, which is particularly valuable for managers and policymakers who want to understand the complete structure of urban and rural green space.
Available via bomb service and API
All improvements are included in the latest version of the NEO tree service. The updated tree data and the new attributes are available through the existing data delivery and API.