ToBRFV-Longitudinal Multispectral Image Dataset

This dataset contains a longitudinal, multispectral image collection created to support early, pre-symptomatic detection of Tomato brown rugose fruit virus (ToBRFV) in two commercial tomato cultivars and two commercial pepper cultivars. A total of 42,736 cropped images were acquired from 347 potted plants (167 tomato, 180 pepper) over a 29-day experimental period (05 September – 03 October 2023). For each plant, up to four images were captured per sampling event across four imaging modalities: standard RGB (converted from DSLR RAW → TIFF), VNIR full-range (400–1100 nm, PNG), VNIR-800 nm (narrowband), and VNIR-1000 nm (narrowband). Imaging was performed from multiple viewing angles using a movable camera platform; DSLR images were captured with a Nikon D90, and VNIR images with an MT-MIMC-V6418 camera. Infection status was confirmed by RT-qPCR at selected time points.

Files are organized into four top-level folders (RGB, VNIR, VNIR_800nm, VNIR_1000nm). Each modality folder contains subfolders for cultivar → health status (healthy / virus-infected) → per-pot image files. Filenames follow a consistent labeling scheme that encodes infection status, day, camera/modality, angle, and pot ID to facilitate automated parsing. Preprocessing steps include visual quality control (removal of overexposed/blurry frames), RAW→TIFF conversion (DCRAW), cropping to plant ROI, and standard filename annotation.