← Marketplace
Tree detection
Description
An algorithm that detects trees in SPOT or Pléiades imagery and returns a probability map. Tree Detection identifies all pixels corresponding to trees in satellite images. Tree detection can be used for infrastructure vegetation risk monitoring, urban planning, construction, fire risk estimation, land management, powerline and trainline inspection, and forestry.
Technical information
| Source | SPOT, Pléiades or Pléiades Neo tasking or catalog collections |
| Compatibility | - The STAC item should be a SPOT, Pléiades or Pléiades Neo image of the display radiometric processing level - The STAC item should have been added to storage in 2023 or later |
| Required parameters | - Console: If your chosen input data meets the above specifications you will be able to run your job - API: You will need to input an Output title and the Input data |
| Output data format | GeoTIFF file: maps the probability of each pixel being part of a tree |
| Output | The result will be added to your account as a STAC item in a new STAC collection. You can retrieve the resulting data in one of the following ways: - Open the console, go to Data management → Jobs - Retrieve the results via the API |
Algorithm performance and training data
The model accuracy ranges from 80 to 90% depending on input imagery resolution and location. This is real-world accuracy following benchmarking against helicopter based LIDAR. Benchmarked against human annotators on 10,000+ Pléiades images the model showed an average precision of 91%, recall of 94.5% and an F1 score of 93%. Please note that for seasonal trees, the algorithm works best when the input images have the trees in full foliage. The model aims to focus on medium to large trees, limiting detection confidence for brushes and ignoring high grass.
The solution uses deep learning techniques for computer vision and is constantly improved with subsequent updates; thus far, it has been trained on tens of thousands of semantically annotated images. The model is used as part of the Tree and tree height detection algorithm.
Due to its complexity, this algorithm requires a significant running time. There are no restrictions on image dimensions, but the computing time will increase with the size of the image.
Further information
For more information, please visit the technical documentation.