Application of computer vision systems for estimation of fat content in poultry meat

FTIR-PCA Approach on Raw and Thermally Processed Chicken Lipids Stabilized by Nano-Encapsulation in β-Cyclodextrin

This study evaluated similarities/dissimilarities of raw and processed chicken breast and thigh lipids that were complexed by β-cyclodextrin, using a combined FTIR-PCA technique. Lipid fractions were analyzed as non-complexed and β-cyclodextrin-complexed samples via thermogravimetry, differential scanning calorimetry and ATR-FTIR. The lipid complexation reduced the water content to 7.67-8.33%, in comparison with the β-cyclodextrin hydrate (~14%). The stabilities of the complexes and β-cyclodextrin were almost the same. ATR-FTIR analysis revealed the presence of important bands that corresponded to the C=O groups (1743-1744 cm^-1) in both the non-complexed and nano-encapsulated lipids. Furthermore, the bands that corresponded to the vibrations of double bonds corresponding to the natural/degraded (cis/trans) fatty acids in lipids appeared at 3008-3011 and 938-946 cm^-1, respectively. The main FTIR bands that were involved in the discrimination of raw and processed chicken lipids, and of non-complexed and complexed lipids, were evaluated with PCA. The shifting of specific FTIR band wavenumbers had the highest influence, especially vibrations of the α(1→4) glucosidic bond in β-cyclodextrin for PC₁, and CH_2/3 groups from lipids for PC₂. This first approach on β-cyclodextrin nano-encapsulation of chicken lipids revealed the possibility to stabilize poultry fatty components for further applications in various ingredients for the food industry.

Keypoint Detection for Injury Identification during Turkey Husbandry Using Neural Networks

Injurious pecking against conspecifics is a serious problem in turkey husbandry. Bloody injuries act as a trigger mechanism to induce further pecking, and timely detection and intervention can prevent massive animal welfare impairments and costly losses. Thus, the overarching aim is to develop a camera-based system to monitor the flock and detect injuries using neural networks. In a preliminary study, images of turkeys were annotated by labelling potential injuries. These were used to train a network for injury detection. Here, we applied a keypoint detection model to provide more information on animal position and indicate injury location. Therefore, seven turkey keypoints were defined, and 244 images (showing 7660 birds) were manually annotated. Two state-of-the-art approaches for pose estimation were adjusted, and their results were compared. Subsequently, a better keypoint detection model (HRNet-W48) was combined with the segmentation model for injury detection. For example, individual injuries were classified using “near tail” or “near head” labels. Summarizing, the keypoint detection showed good results and could clearly differentiate between individual animals even in crowded situations.

The application of computer vision systems in meat science and industry - A review

Computer vision systems (CVS) are applied to macro- and microscopic digital photographs captured using digital cameras, ultrasound scanners, computer tomography, and wide-angle imaging cameras. Diverse image acquisition devices make it technically feasible to obtain information about both the external features and internal structures of targeted objects. Attributes measured in CVS can be used to evaluate meat quality. CVS are also used in research related to assessing the composition of animal carcasses, which might help determine the impact of cross-breeding or rearing systems on the quality of meat. The results obtained by the CVS technique also contribute to assessing the impact of technological treatments on the quality of raw and cooked meat. CVS have many positive attributes including objectivity, non-invasiveness, speed, and low cost of analysis and systems are under constant development an improvement. The present review covers computer vision system techniques, stages of measurements, and possibilities for using these to assess carcass and meat quality.

Accurate prediction and genome-wide association analysis of digital intramuscular fat content in longissimus muscle of pigs

Intramuscular fat (IMF) content is a critical indicator of pork quality that affects directly the purchasing desire of consumers. However, to measure IMF content is both laborious and costly, preventing our understanding of its genetic determinants and improvement. In the present study, we constructed an accurate and fast image acquisition and analysis system, to extract and calculate the digital IMF content, the proportion of fat areas in the image (PFAI) of the longissimus muscle of 1709 animals from multiple pig populations. PFAI was highly significantly correlated with marbling scores (MS; 0.95, r²  = 0.90), and also with IMF contents chemically defined for 80 samples (0.79, r²  = 0.63; more accurate than direct analysis between IMF contents and MS). The processing time for one image is only 2.31 s. Genome-wide association analysis on PFAI for all 1709 animals identified 14 suggestive significant SNPs and 1 genome-wide significant SNP. On MS, we identified nine suggestive significant SNPs, and seven of them were also identified in PFAI. Furthermore, the significance (-log P) values of the seven common SNPs are higher in PFAI than in MS. Novel candidate genes of biological importance for IMF content were also discovered. Our imaging systems developed for prediction of digital IMF content is closer to IMF measured by Soxhlet extraction and slightly more accurate than MS. It can achieve fast and high-throughput IMF phenotype, which can be used in improvement of pork quality.

Modeling Salmonella spp. inactivation in chicken meat subjected to isothermal and non-isothermal temperature profiles

Salmonella genus has foodborne pathogen species commonly involved in many outbreaks related to the consumption of chicken meat. Many studies have aimed to model bacterial inactivation as a function of the temperature. Due to the large heterogeneity of the results, a unified description of Salmonella spp. inactivation behavior is hard to establish. In the current study, by evaluating the root mean square errors, mean absolute deviation, and Akaike and Bayesian information criteria, the double Weibull model was considered the most accurate primary model to fit 61 datasets of Salmonella inactivation in chicken meat. Results can be interpreted as if the bacterial population is divided into two subpopulations consisting of one more resistant (2.3% of the total population) and one more sensitive to thermal stress (97.7% of the total population). The thermal sensitivity of the bacteria depends on the fat content of the chicken meat. From an adapted version of the Bigelow secondary model including both temperature and fat content, 90% of the Salmonella population can be inactivated after heating at 60 °C of chicken breast, thigh muscles, wings, and skin during approximately 2.5, 5.0, 9.5, and 57.4 min, respectively. The resulting model was applied to four different non-isothermal temperature profiles regarding Salmonella growth in chicken meat. Model performance for the non-isothermal profiles was evaluated by the acceptable prediction zone concept. Results showed that >80% of the predictions fell in the acceptable prediction zone when the temperature changes smoothly at temperature rates lower than 20 °C/min. Results obtained can be used in risk assessment models regarding contamination with Salmonella spp. in chicken parts with different fat contents.

Predicting Pork Color Scores Using Computer Vision and Support Vector Machine Technology

The objective of this study was to investigate the ability of image color features to predict subjective pork color scores. Subjective and instrumental color were assessed on the bloomed, cross-sectional surface of pork longissimus thoracis et lumborum chops. Images of pork chop samples were acquired using a computer vision system, and 18 image color features (mean and standard deviation of R, G, B, H, S, I, L*, a*, b*) were extracted for inclusion in partial least squares (PLS) and support vector machine (SVM) regression models. For color scores 2, 3, 4, and 5, the accuracies were 50.4, 75.9, 72.4, and 47.3% classified correctly by PLS, respectively, and 70.7, 72.8, 76.7, and 69.7% by SVM, respectively. The overall prediction accuracies of 2 models for pork color scores were 68.3% for PLS and 73.4% for SVM. There was minimal major misclassification of samples (< 0.5%). Image color features isolated through the development of PLS and SVM models, particularly SVM, show potential as a method to predict pork color scores.

Evaluation of fat grains in gothaj sausage using image analysis

Fat is an irreplacable ingredient in the production of sausages and it determines the appearance of the resulting cut to a significant extent. When shopping, consumers choose a traditional product mostly according to its appearance, based onwhat they are used to. Chemical analysis is capable to determine the total fat content in the product, but it cannot accurately describe the shape and size of fat grains which the consumer observes when looking at the product. The size of fat grains considered acceptable by consumers can be determined using sensory analysis or image analysis. In recent years, image analysis has become widely used when examining meat and meat products. Compared to the human eye, image analysis using a computer system is highly effective, since a correctly adjusted computer program is able to evaluate results with lower error rate. The most commonly monitored parameter in meat products is the aforementioned fat. The fat is located in the cut surface of the product in the form of dispersed particles which can be fairly reliably identified based on color differences in the individual parts of the product matrix. The size of the fat grains depends on the input raw material used as well as on the production technology. The present article describes the application of image analysis when evaluating fat grains in the appearance of cut of the Gothaj sausage whose sensory requirements are set by Czech legislation, namely by Decree No. 326/2001 Coll., as amended. The paper evaluates the size of fat mosaic grains in Gothaj sausages from different manufacturers. Fat grains were divided into ten size classes according to various size limits; specifically, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0, 2.5, 5.0, 8.0 and over 8 mm. The upper limit of up to 8 mm in diameter was chosen based on the limit for the size of individual fat grains set by the legislation. This upper limit was not exceeded by any of the products. On the other side the mosaic had the hightest representation of 0.25 mm fat grains.

Transcriptional background of beef marbling - novel genes implicated in intramuscular fat deposition

The purpose of this study was to identify novel marbling-related genes by comparison of the global gene expression in semitendinosus muscle of 15-month-old Limousin (LIM), Holstein-Friesian (HF) and Hereford (HER) bulls. Muscle of LIM was lean with low intramuscular fat (IMF) content (0.53%) unlike the marbled muscles of HER and HF characterized by higher amounts of IMF (1.10 and 0.81%, respectively). The comparison of muscle transcriptional profile between marbled and lean beef revealed significant differences in expression of 144 genes, presumably involved in consecutive stages of adipose tissue development, such as preadipocyte proliferation and differentiation, adipocyte maturation, lipid filling and lipid metabolism leading to increased IMF deposition and marbling development. Correlation coefficients and regression analysis for nine of them (gadd45a, pias3, ccrn4l, diras3, pou5f1, hoxa9, atp2a2 and pim1) validated by real-time qPCR confirmed their moderate-high correlation with IMF% and explained up to 70.5% of the total variability in IMF deposition in the bulls.

Application of hyperspectral imaging in food safety inspection and control: a review

Food safety is a great public concern, and outbreaks of food-borne illnesses can lead to disturbance to the society. Consequently, fast and nondestructive methods are required for sensing the safety situation of produce. As an emerging technology, hyperspectral imaging has been successfully employed in food safety inspection and control. After presenting the fundamentals of hyperspectral imaging, this paper provides a comprehensive review on its application in determination of physical, chemical, and biological contamination on food products. Additionally, other studies, including detecting meat and meat bone in feedstuffs as well as organic residue on food processing equipment, are also reported due to their close relationship with food safety control. With these applications, it can be demonstrated that miscellaneous hyperspectral imaging techniques including near-infrared hyperspectral imaging, fluorescence hyperspectral imaging, and Raman hyperspectral imaging or their combinations are powerful tools for food safety surveillance. Moreover, it is envisaged that hyperspectral imaging can be considered as an alternative technique for conventional methods in realizing inspection automation, leading to the elimination of the occurrence of food safety problems at the utmost.