VegAnn, Vegetation Annotation of multi-crop RGB images acquired under diverse conditions for segmentation

Applying deep learning to images of cropping systems provides new knowledge and insights in research and commercial applications. Semantic segmentation or pixel-wise classification, of RGB images acquired at the ground level, into vegetation and background is a critical step in the estimation of several canopy traits. Current state of the art methodologies based on convolutional neural networks (CNNs) are trained on datasets acquired under controlled or indoor environments. These models are unable to generalize to real-world images and hence need to be fine-tuned using new labelled datasets. This motivated the creation of the VegAnn - Vegetation Annotation - dataset, a collection of 3775 multicrop RGB images acquired for different phenological stages using different systems and platforms in diverse illumination conditions. We anticipate that VegAnn will help improving segmentation algorithm performances, facilitate benchmarking and promote large-scale crop vegetation segmentation research.

Automated summary

Applying deep learning to images of cropping systems offers new knowledge and insights in research and commercial applications. Semantic segmentation or pixel-wise classification of ground level RGB images into vegetation and background is crucial for estimating canopy traits. Current CNN-based methodologies trained on controlled or indoor datasets cannot generalize to real-world images, requiring fine-tuning with new labeled datasets. The creation of the VegAnn dataset, consisting of 3775 multicrop RGB images acquired under diverse illumination conditions, aims to improve segmentation algorithm performance and facilitate benchmarking in large-scale crop vegetation segmentation research.