🤖 AI Summary
This study addresses the challenge of robustly characterizing tree crown phenology, which is hindered by dynamic spectral, textural, and boundary changes throughout the growing season that confound conventional AI models. To overcome this, the work proposes the first application of self-supervised temporal embedding for individual tree crown phenology modeling, leveraging multi-temporal UAV multispectral imagery. The approach integrates contrastive learning, masked image reconstruction, and explicit seasonal time encoding to produce compact 256-dimensional embeddings. By aligning and cropping crown polygons followed by principal component analysis, the model is trained on 5,885 samples from the HeideBench dataset. The resulting embeddings capture meaningful phenological variation, with the first two principal components explaining 25.1% of the total variance and achieving a median top-1 cosine similarity of 0.946 in nearest-neighbor retrieval—significantly outperforming handcrafted features and mean-pooling baselines.
📝 Abstract
Tree crowns are a challenging target for resilient AI because they are not static objects: their spectral response, internal texture, translucency, and apparent boundaries change substantially across the growing season. We develop PhenoEmbed, a self-supervised crown-centric temporal embedding model trained with contrastive and masked reconstruction objectives on HeideBench, an 18-date UAV multispectral time-series benchmark for forest crown phenology in D{ö}lauer Heide. The model treats seasonal crown dynamics as phenological appearance change driven by leaf emergence, canopy closure, senescence, and leaf-off conditions. Segmented tree crown polygons are retained as object anchors to extract aligned crown-centered crops through time, allowing one 256-dimensional vector summarizing seasonal crown appearance to be learned per tree. On 5,885 crop-safe crowns, the exported embeddings show structured low-dimensional organization, with the first two principal components explaining 25.1\% of variance and nearest-neighbor retrieval producing a median top-1 cosine similarity of 0.946. Compared with handcrafted temporal features and a learned mean-pooling baseline, PhenoEmbed yields substantially more compact nearest-neighbor structure, while ablations show that the contrastive loss, masked reconstruction loss, and explicit seasonal time features each affect the structure of the learned embedding space. These results support PhenoEmbed as a reusable forest crown representation learner and motivate future downstream tests of whether such features improve tree-level models under seasonal change.