A Dataset of Thermal Infrared (TIR) and RGB Images of Lettuces, soil moisture data, and Pseudo-coloring RGB Images of the Lettuce’s Stressed Areas
Description
This dataset supports the article entitled "Integrating Thermal Infrared and RGB Imaging for Early Detection of Water Stress in Lettuces with Comparative Analysis of IoT Sensors". This dataset consists of Thermal Infrared (TIR) images with embedded visible-spectrum (RGB) images of two lettuces (Lactuca sativa var. capitata L), which were captured by a handheld dual Thermal Infrared and RGB camera two times per day for six days. The experiment took place in a lab for two lettuces, where one was irrigated normally, and the other was non-irrigated to have severe water stress. The RGB and TIR images are used initially for the lettuce canopy isolation from the background, the detection of the stressed areas over the lettuce's leaves according to stomatal closure, and considering temperature value differences. The stressed areas in the lettuce canopy are annotated with pseudo-coloring on a newly generated RGB image. Another two groups of pseudo-coloring RGB images were generated where key components were excluded from the hybrid framework. In addition, two capacitive soil moisture sensors log the soil moisture values per hour for six days for each lettuce. The dataset contains 26 raw TIR images with embedded RGB images, 26 RGB images with pseudo-coloring where plants' stressed areas exist, aligned, and cropped based on the TIR images' Field of View (FOV) according to our proposed hybrid framework, 26 RGB images with pseudo-coloring where plants' stressed areas exist excluded the wavelet denoising component, 26 RGB images with pseudo-coloring where plants' stressed areas exist excluded the PCA fusion component, and a file in CSV format with soil moisture values per hour for each lettuce.