ForestSemantic_ITC_Change_2D：Individual Tree Crown Change by Imagery

ForestSemantic_ITC_Change_2D dataset provides 245,616 pair high-resolution individual tree crown (ITC) images for change detetion. 72% belongs to change and 28% belongs to no change. The average pixels for each tree ROI is around 2000 × 2000 pixels.

The high-resolution images were collected from 07/10/2022 to 03/11/2023 by UAV platform. Data collection is usually performed between 2 p.m. and 4 p.m., as often as possible once every two days, but sometimes once every three or four days due to weather conditions, such as rain or strong winds, and on average two or three times a week. A total of 85 time points capturing the changes in trees over one year were collected.

There are 68 trees were selected and cropped for change analysis. The trees are in diversity, from evergreen to deciduous, flowering, and fruit-bearing trees. To thoroughly investigate tree changes, we perform a pairwise comparative analysis of tree images at all time points for each tree. In total, we have 

This dataset is the supplement data of our publication inCVPR 2025: Deep Change Monitoring: A Hyperbolic Representative Learning Framework and a Dataset for Long-term Fine-grained Tree Change Detection