Using Deep Convolutional Neural Network for oak acorn viability recognition based on color images of their sections.

•An application of a Neural Network to the recognition of the quality of manually scarified oak seeds.•A method of assessing the sowing viability of acorns and automation of their scarification.•A description of acquires images of seed sections, their preprocessing and augmentation.•Details of the modified CNN Deep Neural Network architectures and their training.•Experimental results and comparison of several Deep Network models and augmentation schemes.•Analysis of an impact of representation of colour space and topography of sections on recognition.•Discussion about what the network learns from the training data.