The scientific project: "Development of a method and device for assessing the sexual dimorphism of poultry egg embryos based on computer vision methods and artificial neural networks" is being implemented under the leadership of Alexander Aleinikov, head researcher of SFSCA RAS.
The study was funded by the grant No. 22-26-00198 from the Russian Science Foundation.
Poultry eggs contain a large number of valuable micronutrients and nutrients. They are the main source of protein and are used to prepare a large number of traditional dishes, as well as they are indispensable for dietary and sports nutrition. There is a gender preference in poultry meat production, where the male is preferred in broiler production. However, roosters cannot be used in egg production. [https://blog.minifermer.org›read/17/19-tehnologiya-..].
For broiler production, female chicks are inferior to males because of their lower growth rate compared to their peers. Female chicks require additional feed costs to gain weight.
Furthermore, female chicks in broiler farms are not economically justified when sent to market because of lower body weight gain compared to male chicks. Therefore, more than 7.0 billion day-old chicks hatched from eggs are killed each year worldwide by carbon dioxide asphyxiation or maceration because of this preference for gender-oriented production [https://ru.wikibrief.org›wiki/Chick_culling..]. This attitude toward live chickens causes serious ethical problems and results in significant economic losses.
Many researchers try to apply different strategies to determine the sex of an embryo before the chicks leave the egg based on differences in the DNA content of the blastoderm and hormonal differences in the allantois fluid of embryos. But these methods are rarely used in practice because they destroy the integrity of the egg structure.
The methods of computer vision application during incubation by analyzing images of development of blood vessels of embryos, as they are supported by a significant theoretical basis are rather promising. Computer vision systems can provide reliable, non-destructive and precision technology for probing and monitoring various aspects of poultry production processes. They will provide the formation of a large variety of data to assess the viability and prognosis of sexual characteristics of egg embryos for subsequent analysis.
However, to follow the path of foreign machine learning research conducted in 2021-2022 based on the created genetic algorithm to optimize the initial weights and thresholds of eleven blood vessel traits that characterize the sexual dimorphism of embryos in the first half of incubation [https://sci-hub.ru doi: 10.1016/j.japr.100203], is not rational.
Therefore, based on computer vision, new methods have been proposed to determine sexual dimorphism during the initial days of incubation (days 1-3), namely:
1) the method of determining sexual dimorphism by detecting microvessels of blood vessels and determining their size distribution in a selected local area on the image of an egg yolk, including blastodisc. An algorithm and a computer program were developed to carry out research in this direction, designed to predict dimorphism in the first days based on the difference of microvessel distribution according to the constructed histogram of discrete particle number values by computer vision method.
The method has legal protection - the certificate of registration of computer program RU № 2022664104 Computer program "Determination of the number and distribution of microparticles in the image of the surface of the object", published in FIIP 22.07.2022;
2) the method for determining sexual dimorphism, according to the difference in the dynamics of embryonic heartbeats during incubation. Given that cardiac activity in the embryo is detected on the 2nd day of incubation, it is possible to develop a new method for early assessment of sexual dimorphism in the incubated egg, based on the hypothesis of differences in the frequency and dynamics of heart contractions in rooster and hen embryos. Such a method for assessing sexual dimorphism was not found in the analyzed sources. The fact of differences in the cardiac activity of the two sexes of embryos can be refuted or proved only by applying modern methods and computer vision tools. If, in the future, such an approach cannot be implemented, in this case it can be used in controlling the viability of embryos during incubation.
To confirm the possibility of implementing new methods and science-based approaches, a new multifunctional research complex of technical means for the implementation of the proposed methods and approaches was developed, including:
а) calorimetric apparatus for obtaining RGB-images by computer vision;
б) devices for the study of hyperspectral spectrum from the surface of the object of the study;
в) impedance spectroscopy facilities for studying the electrical parameters of the research object;
г) facilities for assessing viability and creating the necessary conditions for incubation and hatching chicks.
The use of this complex in the research will make it possible in the future to conduct large-scale research on the problem being solved and to obtain reliable results on the choice and evaluation of the effectiveness of the proposed methods and approaches to determine the sexual dimorphism of embryos in the pre-incubation and incubation period.
The list of scientists who are working on this project:
1) Alexander Fedorovich Aleinikov, Head Researcher SFSCA RAS (project manager and executor);
2) Evgeny Sergeevich Smirnov, Researcher, Technological Design Institute of Scientific Instrument Engineering SB RAS (project executor);
3) Iliya Sergeevich Dubkov, senior lecturer of the NSTU Department of Computer Engineering (project executor);
4) Ivan Viktorovich Osipenko, postgraduate student of the NSTU Department of Computer Engineering (project executor).