Individual Tree Crown Study by Imagery

The ForestSemantic_ITD_2D includes a significant amount of annotated Individual Tree Delineation (ITD) reference images, are the result of a collaborative effort from numerous contributors worldwide. 

The dataset covers 11 study sites across 9 countries including Australia, Canada, China, Germany, Kenya, Malaysia, Norway, Panama, and United States, spanning tropical, subtropical, and temperate climate zones. The dataset represents diverse forest types, such as tropical rainforests, montane forests, moist forests, and savanna woodlands; subtropical evergreen broadleaf and mixed forests; and temperate broadleaf and mixed forests. In addition, the dataset also includes both natural and urban forests.

High-resolution images were acquired from the study sites using airplane or UAV platforms, with resolutions ranging from 2 cm to 10 cm. The original images were then cropped into individual image tiles with a size of 1024×1024 pixels for further processing. Table 1 lists detailed information about the study sites and the conditions of their corresponding dataset.

As the reference for the training and evaluation, 600,000 manually annotated ITC masks were provided for over 11,000 images. Annotations of the visible dominant and co-dominant ITC on images were conducted through visual interpretation.

This dataset supports the ITD Contest, 2024 and a bechmark study. This dataset will be public soon.

Figure 1. The distribution of study site and sample data.