Increasing accuracy in classification models for the identification of plant species based on UAV images

The identification of plant species from RGB images is a challenge of growing importance in the field of biodiversity assessment. This study aims to develop an image pre-processing procedure that increases the accuracy of classification models applied to low-resolution plant images collected by RGB Unmanned aerial vehicles (UAVs). This procedure, based on contrast enhancement and superresolution techniques, has been successfully tested on RGB images collected in agroecosystems.