"A semi-supervised framework for UAV-based individual tree crown segmentation in structurally heterogeneous planted forests"
(Qing Wang, Yihui Zhao, Yingpu Che, Han Shen, Yongbin Qiu, Yixiang Wang. [International Journal of Applied Earth Observation and Geoinformation] [2026])
We pioneer a semi-supervised learning (SSL) framework that synergistically integrates unsupervised pseudo-label generation, Mask R-CNN-based instance segmentation, and staged fine-tuning strategy.