SPAD502 Plus instruments and tape measures were used to collect SPAD and cotton plant height data, revealing a strong positive correlation between cotton height and SPAD throughout the growth period.

The study emphasizes the importance of accurately monitoring chlorophyll content as an indicator of crop health, especially in intercropping systems.

Researchers utilized unmanned aerial vehicles (UAVs) to collect visible and multispectral data during three growth stages of cotton in the Yellow River Delta region of China.

This technology addresses the limitations of traditional manual sampling methods, providing broader coverage and insights into crop health across large areas.

The study proposes an automated workflow for single-tree segmentation using high-resolution multispectral UAV images, aimed at enhancing forest management practices.

To achieve effective segmentation, a color thresholding method in the L*a*b* color space was employed, which worked well for healthy trees but faced challenges with phytosanitary issues.

Among the deep learning models tested, U-Net achieved the best performance, recording an F1-score of 0.56 and a counting accuracy of 0.71 using a combination of two datasets for training.

Faster R-CNN also showed strong results, achieving an F1-Score of 83.5% and an Intersection over Union (IoU) of 65.3% when trained on specific spectral bands.

The database used for these analyses consisted of multispectral UAV data from a forest area in Germany, featuring eight tree species in a matured closed canopy stage.

Previous research primarily focused on image metrics, but this study highlights the potential benefits of integrating DAP point cloud metrics for biomass estimation.

The research findings suggest that integrating multi-height UAV imagery with DAP metrics is promising for enhancing biomass estimation in Ginkgo saplings and potentially other tree crops.

The TreeSeg toolbox, developed from the study, allows users to perform instance segmentation of individual trees in multispectral images, with the models available on GitHub.