AI-BASED MODEL FOR IDENTIFYING LARGE ASPENS FROM AERIAL IMAGES

Based on an artificial intelligence model, the method developed by KOKO Forest Oy for identifying aspen (Populus tremula L.) from aerial images was found to be reliable, particularly for large aspens. A recently completed thesis investigated the reliability of this method, focusing on forests dominated by pine and spruce. Aspen is an important tree species for biodiversity, and understanding its distribution supports sustainable forest management. Species diversity is especially rich in large aspens.

The study analyzed results produced by artificial intelligence using terrain data from two different areas. The AI reliably identified aspens with a diameter at breast height greater than 30 cm in the upper canopy layer. However, smaller aspens growing in lower canopy layers cannot be visually detected from aerial images, making their identification with AI models challenging. The method identified approximately half of the aspens in the test areas, with a detection of 71% for large aspens. Unrecognized were, for example, aspens located by the roadside, as they did not stand out sufficiently from their surroundings.

The study demonstrated that the data produced by this method can already be utilized to assess forest ecosystem diversity and support forest management planning.

Read more about the thesis conducted at Häme University of Applied Sciences (in Finnish with English abstract): Haapojen tunnistus ilmakuvista tekoälyä hyödyntäen