Plant phenotyping based on multimodal time-series data: 3D reconstruction, point cloud segmentation, and yield prediction

In this project, I focus on multi-species plant 3D point clouds as the research object.Through the extraction and analysis of various phenotypic parameters from point clouds, the goal is to predict the growth, structure, and yield of plants. The specific objectives are as follows. 1) Point cloud reconstruction algorithm for individual and population of greenhouse crops. 2) Instance segmentation model for individual crops and semantic segmentation model for crop populations. 3) Using point cloud time-series data and multi-source sensor data in the greenhouse, establish a functional structural model to predict yield and phenotypic traits. 4) Build a well-labelled plant point cloud dataset.