Douglas Fernandes Barbin

Predicting the ripening of papaya fruit with digital imaging and random forests

Abstract Papaya grading is performed manually which may lead to misclassifications, resulting in fruit boxes with different maturity stages. The objective is to predict the ripening of the papaya fruit using digital imaging and random forests. A series of physical/chemical analyses are carried out and true maturity stage is derived from pulp firmness measurements. Imaging and image analysis provides hand-crafted color features computed from the peel and random decision forests are implemented to predict ripening stage. More specifically, a total of 114 samples from 57 fruits are used for the experiments, and classified into three stages of maturity. After image acquisition and analysis, twenty-one hand-crafted color features (comprising seven groups) that have low computational cost are extracted and evaluated. Random forests with two datasets (cross-validation and prediction set) are employed for the experiments. Concerning all image features, 94.3% classification performance is obtained over the cross-validation set. The prediction set obtained 94.7% misclassifying only a single sample. For the group comparisons, the normalized mean of the RGB (red, green, blue) color space achieved better performance (78.1%). Essentially, the technique can mature into an industrial application with the right integration framework.

Processo de congelamento em túnel portátil com convecção forçada por exaustão e insuflação para paletes

The objective of this work is to build an experimental portable forced-air freezing tunnel which creates a low or high pressure region surrounding the product. Comparative studies with air exhausting and blowing were conducted. The tunnel was built and placed inside a freezing product storage chamber, and the objective was to improve the air circulation and the thermal distribution between the product and cold air for a sample batch left inside the chamber. A heat transfer analysis comparing the process and the air distribution around the product was performed. The air evacuation process reduced up to 14% of the freezing time in relation to the blowing system and 10% in relation to the mixed system.

VIS–NIR spectroscopy as a process analytical technology for compositional characterization of film biopolymers and correlation with their mechanical properties

There is an increasing interest in the use of polysaccharides and proteins for the production of biodegradable films. Visible and near-infrared (VIS-NIR) spectroscopy is a reliable analytical tool for objective analyses of biological sample attributes. The objective is to investigate the potential of VIS-NIR spectroscopy as a process analytical technology for compositional characterization of biodegradable materials and correlation to their mechanical properties. Biofilms were produced by single-screw extrusion with different combinations of polybutylene adipate-co-terephthalate, whole oat flour, glycerol, magnesium stearate, and citric acid. Spectral data were recorded in the range of 400-2498nm at 2nm intervals. Partial least square regression was used to investigate the correlation between spectral information and mechanical properties. Results show that spectral information is influenced by the major constituent components, as they are clustered according to polybutylene adipate-co-terephthalate content. Results for regression models using the spectral information as predictor of tensile properties achieved satisfactory results, with coefficients of prediction (R(2)C) of 0.83, 0.88 and 0.92 (calibration models) for elongation, tensile strength, and Youngs modulus, respectively. Results corroborate the correlation of NIR spectra with tensile properties, showing that NIR spectroscopy has potential as a rapid analytical technology for non-destructive assessment of the mechanical properties of the films.

Machine Learning Applied to Near-Infrared Spectra for Chicken Meat Classification

Identification of chicken quality parameters is often inconsistent, time-consuming, and laborious. Near-infrared (NIR) spectroscopy has been used as a powerful tool for food quality assessment. However, the near-infrared (NIR) spectra comprise a large number of redundant information. Determining wavelengths relevance and selecting subsets for classification and prediction models are mandatory for the development of multispectral systems. A combination of both attribute and wavelength selection for NIR spectral information of chicken meat samples was investigated. Decision Trees and Decision Table predictors exploit these optimal wavelengths for classification tasks according to different quality grades of poultry meat. The proposed methodology was conducted with a support vector machine algorithm (SVM) to compare the precision of the proposed model. Experiments were performed on NIR spectral information (1050 wavelengths), colour (CIE , chroma, and hue), water holding capacity (WHC), and pH of each sample analyzed. Results show that the best method was the REPTree based on 12 wavelengths, allowing for classification of poultry samples according to quality grades with precision. The selected wavelengths could lead to potential simple multispectral acquisition devices.

Classification and compositional characterization of different varieties of cocoa beans by near infrared spectroscopy and multivariate statistical analyses

Abstract Effective and fast methods are important for distinguishing cocoa varieties in the field and in the processing industry. This work proposes the application of NIR spectroscopy as a potential analytical method to classify different varieties and predict the chemical composition of cocoa. Chemical composition and colour features were determined by traditional methods and then related with the spectral information by partial least-squares regression. Several mathematical pre-processing methods including first and second derivatives, standard normal variate and multiplicative scatter correction were applied to study the influence of spectral variations. The results of chemical composition analysis and colourimetric measurements show significant differences between varieties. NIR spectra of samples exhibited characteristic profiles for each variety and principal component analysis showed different varieties in according to spectral features.

Fuzzy approach for classification of pork into quality grades: coping with unclassifiable samples

Abstract Meat classification methods are commonly based on quality parameters standardized by numeric ranges. However, some animal samples from different production chains do not match the current grades proposed. These unclassifiable samples are not capable to fit into a standard created by crisp range of values due to being infeasible toward its definition. An alternative to handle this kind of sample classification is the fuzzy logic, which could deal with uncertainty and ambiguity degree like human reasoning. In this work, we compare the traditional classification method and fuzzy approaches with the objective to handle the infeasible samples. This was compared to traditional pork standards using eleven real-life datasets with a total of 1798 samples described by pH, water holding capacity and/or L∗ value. The results demonstrated that traditional classification could not predict the unclassifiable samples. On the other hand, the fuzzy approaches improve significantly the number of classified samples. Performance of the fuzzy approaches were compared with several machine learning algorithms, but no significant statistical difference was observed. Finally, a real-life study case was explored, highlighting some advantages and further achievements of the fuzzy modeling.

Classification of Chicken Parts Using a Portable Near-Infrared (NIR) Spectrophotometer and Machine Learning

Identification of different chicken parts using portable equipment could provide useful information for the processing industry and also for authentication purposes. Traditionally, physical–chemical analysis could deal with this task, but some disadvantages arise such as time constraints and requirements of chemicals. Recently, near-infrared (NIR) spectroscopy and machine learning (ML) techniques have been widely used to obtain a rapid, noninvasive, and precise characterization of biological samples. This study aims at classifying chicken parts (breasts, thighs, and drumstick) using portable NIR equipment combined with ML algorithms. Physical and chemical attributes (pH and L*a*b* color features) and chemical composition (protein, fat, moisture, and ash) were determined for each sample. Spectral information was acquired using a portable NIR spectrophotometer within the range 900–1700 nm and principal component analysis was used as screening approach. Support vector machine and random forest algorithms were compared for chicken meat classification. Results confirmed the possibility of differentiating breast samples from thighs and drumstick with 98.8% accuracy. The results showed the potential of using a NIR portable spectrophotometer combined with a ML approach for differentiation of chicken parts in the processing industry.

Caracterização tecnológica de massa seca tipo Fettucine, adicionada de diferentes tipos de fibras

Resumo As massas secas alimentícias estão presentes na mesa do consumidor brasileiro, fazendo parte das principais refeições em todas as classes socias. Entretanto, a crescente busca por alimentos mais saudáveis têm levado a indústria a alterar a formulação de vários produtos, com a adição de fibras e/ou proteínas, ou redução do teor calórico (açúcar e gordura). Assim, avaliamos a adição de diferentes tipos de fibras em formulações de massas alimentícias secas, tipo fettucine, com o objetivo de aumentar os efeitos benéficos à saúde do consumidor.

Implications of Non-Equilibrium States and Glass Transitions in Frozen and Dried Fish and Meat Products

Abstract Although frozen and dried foods are microbiologically stable, they are susceptible to chemical and physical deterioration during storage, such as enzymatic changes, oxidation, and crystallization of solutes affecting product quality. The desiccated food could be stored in an amorphous glassy state for improved long-term stability. However, the amorphous product may undergo structural and chemical changes if storage temperature is above the glass-transition temperature. Hence, determination and understanding of glass-transition temperature is especially important in stability studies of food matrices, in order to determine the best storage or processing condition for a frozen or dried product. In this chapter, some general aspects of freezing and drying processes of fish and meat are presented. Some works where glass-transition approach was used to explain chemical and physical changes of frozen and dried fish and meat during storage are described.

Computer vison for classification and quality analysis of food.

It was investigated the classification of diferent kinds of pasta according to their composition by image analyses