AUTOMATED NON-INVASIVE ASSESSMENT OF POULTRY FERTILIZATION Russian patent published in 2020 - IPC G01N33/08 G01N24/08 G01R33/483 

FIELD: agriculture.

SUBSTANCE: group of inventions relates to automated non-invasive determination of poultry fertilization. Method comprises the following steps: sequential or parallel feeding a plurality of poultry eggs into an NMR apparatus, exposing poultry eggs to NMR measurement, for example, for generation of three-dimensional NMR image of at least part of each of said eggs, wherein said three-dimensional NMR image has spatial resolution in at least one measurement of 1.0 mm or less, preferably - 0.50 mm or less, wherein said egg part (14) comprises a germ disc of the corresponding egg, determining a fertilization prediction according to at least one of the following two procedures: (i) detecting at least one feature from each of said three-dimensional NMR images and using said at least one feature in a classifier based on the criteria for determining a fertility prediction, and (ii) using an in-depth training algorithm, and in particular an in-depth training algorithm based on convolutional neural networks, generative-adversarial networks, recurrent neural networks or neuron long-term memory networks.

EFFECT: application of this group of inventions enables to treat a large number of eggs without damaging or changing eggs in any way.

37 cl, 16 dwg