Synthetic HDR dataset of RGB-D images of plants and their masks for semantic and instance segmentation

With the need of a large amount of data in the fields of Deep Learning and Computer Vision to train Neural Networks to either classify images, locate objects or segment instances and also the lack of real data, synthetic data became a necessity. Nowdays, we train our ML/DL models with synthetic data then we use transfert learning with the few available real data. This dataset was used for semantic and instance segmentation of plant. I used a CNN model to predict the class of each pixel ( fruit, stem, petiole, leaf or background ) than with the depth mask I measured the instances dimensions.