Maize-CIPT: A High-Frequency, Dual-Modal Imagery Time-Series Dataset Coupled with Canopy-Individual Agronomic Traits for Maize Phenology

Accurate monitoring of maize phenology is critical for precision agriculture and food security, yet remains constrained by the lack of long-term, high-frequency observations. Here we present Maize-CIPT (High-frequency, Canopy–Individual Coupled, Dual-modal Imagery), a time-series dataset integrating near-surface imagery with in-situ agronomic and phenological measurements across two summer maize growing seasons. Data were acquired using a PTZ-controlled NDVI camera, enabling observation of two spatially separated plots for independent data partitioning, and near-synchronous acquisition with native spatial co-registration of RGB and NDVI imagery via automated filter switching. Synchronized ground-truth measurements include plant height, leaf age, leaf inclination angle, and phenological stage. Following rigorous cleaning and spatiotemporal alignment, the dataset comprises 2,184 high-quality records. Data reliability was systematically validated, and the value of high-frequency observations was assessed through downsampling experiments. Maize-CIPT supports the development and benchmarking of computer vision and machine learning methods for applications including real-time phenology monitoring and high-throughput crop phenotyping, and enables further data expansion through flexible image cropping and fine-grained annotation.