Nondestructive identification for gender of chicken eggs based on GA-BPNN with double hidden layers

A review of the recent advances for the in ovo sexing of chicken embryos using optical sensing techniques

The culling of day-old male chicks has caused ethical and economic concerns. Traditional approaches for detecting the in ovo sex of chicken embryos involve opening the eggshell and inner membrane, which are destructive, time-consuming, and inefficient. Therefore, noncontact optical sensing techniques have been examined for the in ovo sexing of chicken embryos. Compared with traditional methods, optical sensing can increase determination throughput and frequency for the rapid sexing of chicken embryos. This paper presented a comprehensive review of the different optical sensing techniques used for the in ovo sexing of chicken embryos, including visible and near-infrared (Vis-NIR) spectroscopy, hyperspectral imaging, Raman spectroscopy, fluorescence spectroscopy, and machine vision, discussing their advantages and disadvantages. In addition, the latest research regarding different detection algorithms and models for the in ovo sexing of chicken embryos was summarized. Therefore, this paper provides updated information regarding the optical sensing techniques that can be used in the poultry industry and related research.

Trends in in ovo sexing technologies: insights and interpretation from papers and patents

Numerous researchers and institutions have been developing in ovo sexing technologies to improve animal welfare by identifying male embryos in an early embryonic stage and disposing of them before pain perception. This review gives a complete overview of the technological approaches reported in papers and patents by performing a thorough search using Web of Science and Patstat/Espacenet databases for papers and patents, respectively. Based on a total of 49 papers and 115 patent families reported until May 2023 worldwide, 11 technology categories were defined: 6 non-optical and 5 optical techniques. Every category was described for its characteristics while assessing its potential for application. Next, the dynamics of the publications of in ovo sexing techniques in both paper and patent fields were described through growth curves, and the interest or actual status was visualized using the number of paper citations and the actual legal status of the patents. When comparing the reported technologies in papers to those in patents, scientific gaps were observed, as some of the patented technologies were not reported in the scientific literature, e.g., ion mobility and mass spectrometry approaches. Generally, more diverse approaches in all categories were found in patents, although they do require more scientific evidence through papers or industrial adoption to prove their robustness. Moreover, although there is a recent trend for non-invasive techniques, invasive methods like analyzing DNA through PCR or hormones through immunosensing are still being reported (and might continue to be) in papers and patents. It was also observed that none of the technologies complies with all the industry requirements, although 5 companies already entered the market. On the one hand, more research and harmony between consumers, industry, and governments is necessary. On the other hand, close monitoring of the market performance of the currently available techniques will offer valuable insights into the potential and expectations of in ovo sexing techniques in the poultry industry.

Single-View Measurement Method for Egg Size Based on Small-Batch Images

Egg size is a crucial indicator for consumer evaluation and quality grading. The main goal of this study is to measure eggs' major and minor axes based on deep learning and single-view metrology. In this paper, we designed an egg-carrying component to obtain the actual outline of eggs. The Segformer algorithm was used to segment egg images in small batches. This study proposes a single-view measurement method suitable for eggs. Experimental results verified that the Segformer could obtain high segmentation accuracy for egg images in small batches. The mean intersection over union of the segmentation model was 96.15%, and the mean pixel accuracy was 97.17%. The R-squared was 0.969 (for the long axis) and 0.926 (for the short axis), obtained through the egg single-view measurement method proposed in this paper.

An approach for goose egg recognition for robot picking based on deep learning

1. In a non-cage environment, goose eggs are buried in litter and goose feathers, leading to contamination and discolouration. Such random distribution of goose eggs poses a great challenge to the recognition and location of intelligent picking by robot systems on farm.2. In order to assist in recognition and location of goose eggs in non-cage environment, a novel method was proposed which used three-channel convolutional neural network (T-CNN), composed of improved AlexNet, combined with 'you only look once' (YOLOv5), egg contour curve creation and support vector machine (SVM).3. Using this method, the original goose egg images were inputted into the YOLOv5 model for target detection and segmentation. In parallel, the median filter and maximum interclass variance method (OTSU) were applied to egg segmentation images to obtain the main pixels for each, and the Kirsch operator was used for edge extraction and contour curves fitting by designing the fitting curve equation to obtain segmentation images with goose egg contour curves.4. In order to further enrich the differences between goose eggs and background, the goose egg segmentation images were divided into three colour components: R, G and B, which were input into T-CNN for feature extraction. Then the eggs were classified by vector stitching and SVM, by adding goose egg contour curves images.5. The recognition and location results showed that about 95.65% of the goose egg pixel blocks in the segmented images are recognised correctly. About 3.81% of the pixel blocks in the segmented images were recognised incorrectly, and the centre of mass offset was about 4.45 pixels.6. This study demonstrated accurate goose egg recognition and location effects using the proposed method in a non-cage environment This highlighted its application prospect in intelligent goose eggs picking.

Artificial neural network-genetic algorithm-based optimization of aerobic composting process parameters of Ganoderma lucidum residue

The rapid development of traditional Chinese medicine enterprises has put forward higher requirements for the resource utilization of traditional Chinese medicine residues (TCMR). Aerobic composting of TCMR to prepare bio-organic fertilizer is an effective resource utilization method. In this study, a back-propagation artificial neural network (BPNN) model using composting factors as inputs (C/N, initial moisture content, type of inoculant, composting days) and the humic acid content as the output was constructed based on the orthogonal test data. BPNN-GA (a genetic algorithm) was used for extreme value optimization, and the optimal composting process parameter combination was obtained and verified. The results show that the combination of orthogonal testing and BPNN can effectively establish the relationship between the composting process parameters and humic acid content. The R² value was 0. 9064. The optimized parameter combination is as follows: C/N,37.42; moisture content,69.76%; bacteria,no; and composting time,50 d.

Artificial Intelligence Network Embedding, Entrepreneurial Intention, and Behavior Analysis for College Students' Rural Tourism Entrepreneurship

To promote the development of the rural economy and improve entrepreneurship education in colleges and universities, college students’ willingness and behavior toward rural tourism entrepreneurship were investigated in this study. First of all, based on the previous research results, the influencing factor model was determined for college students’ entrepreneurial intention. Second, a questionnaire survey was made to collect data from a university in Xi’an City. Finally, the artificial neural network (ANN), improved by a genetic algorithm (GA) based on an artificial intelligence network, was used to study the relationship between college students’ entrepreneurial intention and behavior, and the simulation was carried out on MATLAB2013b software. The results show that the average evaluation accuracy is 81.13% for 60 groups of data using the unmodified back propagation neural network (BPNN) algorithm, while the average evaluation accuracy is 92.17% for the BPNN algorithm improved and optimized by GA, with an ascent of 11.04%. Therefore, the BPNN algorithm improved and optimized by GA is better than the unmodified BPNN algorithm; It is also feasible and effective in the analysis of influencing factors of college students’ entrepreneurial intention and behavior. The research provides a basis for colleges and universities to carry out entrepreneurship education on a large scale and to cultivate their innovative talents.

Risk Assessment and Prediction of Air Pollution Disasters in Four Chinese Regions

Evaluating the regional trends of air pollution disaster risk in areas of heavy industry and economically developed cities is vital for regional sustainable development. Until now, previous studies have mainly adopted a traditional weighted comprehensive evaluation method to analyze the air pollution disaster risk. This research has integrated principal component analysis (PCA), a genetic algorithm (GA) and a backpropagation (BP) neural network to evaluate the regional disaster risk. Hazard risk, hazard-laden environment sensitivity, hazard-bearing body vulnerability and disaster resilience were used to measure the degree of disaster risk. The main findings were: (1) the air pollution disaster risk index of Liaoning Province, Beijing, Shanghai and Guangdong Province increased year by year from 2010 to 2019; (2) the mean absolute error (MAE), root mean square error (RMSE) and mean absolute percentage error (MAPE) of each regional air pollution disaster risk index in 2019, as predicted by the PCA-GA-BP neural network, were 0.607, 0.317 and 20.3%, respectively; (3) the predicted results were more accurate than those using a PCA-BP neural network, GA-BP neural network, traditional BP neural network, support vector regression (SVR) or extreme gradient boosting (XGBoost), which verified that machine learning could be used as a method of air pollution disaster risk assessment to a considerable extent.