Max David Quispe Bonilla

The classification of alpaca fleeces is a crucial process in the textile value chain, directly impacting the price and final quality of the products. The Peruvian Technical Standard 231.301:2022 establishes quality groups for fibre classification. Traditionally, this task is performed by expert classifiers («master graders») using sight and touch, a laborious method susceptible to variability. Given the need to optimize this process, the use of artificial intelligence(AI)-based technology has emerged. Therefore, the objective of this research was to compare the accuracy, productivity, and economic viability of three methods for classifying alpaca fibre: an automatic AI-based classifier (Fiber Class), objective classification using measurements with an electronic fibre characterizer (considered the «Standard»), and the traditional subjective method («master grader»). Twenty-one fleece samples previously classified by the master graders were used, covering seven quality groups. Each sample classified by the «master grader» was evaluated by the Fiber Class and then reclassified using the «Standard» method, subdividing them and measuring each mini sample with the Fibre EC to determine the actual quality and internal proportions. Productivity was evaluated by measuring the time each method takes to classify alpaca fibre fleece samples (kg/month); an economic analysis was also performed. Fiber Class achieved an accuracy of 85.7 ± 7.6%, with no significant difference from the «Standard» method (p>0.05), while the «master graders» obtained 47.6 ± 7.6% accuracy, with a tendency to overestimate quality, being significantly different (p<0.01) from the «Standard» method. Fiber Class also provided information on the proportion of different qualities within each sample and, in terms of productivity (3463 kg/month), surpassed the «master graders» (2304 kg/month) and the «Standard» method (12 kg/ month). Economically, Fiber Class generated the highest monthly profit due to its high productivity and lower cost per kilogram, despite higher total costs, while the «Standard» method proved unviable due to its high cost. The findings demonstrate the potential of Fiber Class to improve objectivity, accuracy, efficiency, and profitability in alpaca fibre classification.

Alpaca fiber is classified as a special elite or luxury fiber, characterized by its exceptional comfort, softness, good insulation, lightness, lustre, durability, hypoallergenic, breathability, hea...

Alpaca fibre exhibits different quality grades based on fibre diameter and is regulated by the Peruvian Technical Standard – NTP 231.301.2022; however, there are few studies on its cross-sectional characteristics and tensile strength as a function of these different qualities. This study aimed to investigate the cross-sectional characteristics, fibre diameter, and tensile strength in relation to the quality grades. Seven fibre quality groups were established by measuring fibre bundles, and four samples were taken from each group. The cross-sectional characteristics, fibre diameter, and tensile strength were measured using the Fiber Mult, Fiber EC, and Fiber TST devices, respectively. All three variables were affected by fibre quality (p<0.05), showing larger cross-sectional areas and diameters in lower-quality fibres, while tensile strength increased up to quality grade 5 (Fine) before beginning to decrease, possibly due to the higher incidence of medullated fibres and larger medulla area in the lower-quality groups. On the other hand, the tensile strength is higher than that of wool fibres and would not present problems during textile processing. High and positive correlations were obtained between the mean fibre diameter (MFD) and the cross-sectional characteristics, equivalent mean fibre diameter (MFD_e), and apparent mean fibre diameter (MFD_a). The cross-sectional characteristics of alpaca fibre are good estimators of the MFD, but the MFD_e, MFD_a, and cross-sectional characteristics would be the best indicators of fibre fineness.

The fleeces of mammals with dense coats, such as the mohair fleeces of Angora goats, usually include medullated fibres. These fibres constitute a problem for the textile industry because of their structural characteristics. Three experiments were conducted in this study, with the aim of comparing human image analysis to digital image analysis and artificial intelligence (AI), in terms of their ability to determine the incidence of medullation in mohair samples. The experiments entailed determining the incidences of industry non-objectionable medullated (NOB) fibres and objectionable medullated (SME) fibres, as percentages of the non-medullated fibres. In each experiment, a set of samples was analysed by both laboratory personnel and by different AI models using a Smart Fiber Medullometer. Laboratory personnel showed better coincidence and higher correlations with the AI models when counting SME fibres (r = 0.64–0.97) than when counting NOB fibres (r = 0.57–0.87). This could be the result of the more clearly defined characteristics of SME fibres, in relation to NOB fibres. The results of this studyindicate a great advance in the automatic detection of SME and NOB fibres in mohair samples. However, further adjustments of the AI models are required for counting NOB fibres.

Animal-origin fibers exhibit characteristics that determine their quality, price, and comfort, with diameter and medullation type being particularly important. Medullated fibers are considered problematic for the textile industry as they increase the sensation of itchiness on the skin and the heterogeneity of fabrics, leading to a decrease in fiber quality. Thus, having practical instruments to identify them is very essential. On the other hand, some fibers, depending on the species they come from, are considerably cheaper than others. Hence the existence of counterfeit products, which are detrimental to legitimate producers and consumers alike. Fourier-Transform Infrared (FTIR) spectroscopy can extract a characteristic waveform from fibers than can be later used for classification. However, visually inspecting such waveforms is imprecise. In parallel, Artificial Intelligence is an emerging field that could be useful in this area, offering new possibilities and significant advances. Therefore, the first aim of this thesis was to develop and validate a novel automatic system (referred to as S-Fiber Med) for medullation and diameter assessment of animal fibers based on artificial intelligence. The second objective was to develop and validate two machine learning models, namely, Deep Neural Networks (DNN) and Support Vector Machine (SVM) to classify spectra of fibers by species. For the first objective, the medullation of 88 samples of alpaca, llama and mohair fibers (41, 43 and 4, respectively) was evaluated. Additionally, 269 samples of alpacas were considered for average fiber diameter (AFD) and the results were compared with the Portable Fiber Tester (PFT) and Optical Fibre Diameter Analyser (OFDA) methods (72 and 197 samples, respectively). For the second objective, fiber spectra were acquired using an FTIR spectrometer in Attenuated Total Reflectance (ATR) mode (FTIR-ATR). Camelid (alpaca: n = 51, llama: n=50, vicuña: n = 50) and goat (mohair: n = 35 and cashmere: n = 20) samples were evaluated, from which 1,236 FTIR-ATR spectra were obtained. Results of our first objective indicate that there was no significant difference between medullation percentage obtained with the projection microscope and the S-Fiber Med. The Pearson and Concordance correlation analysis shows a strong, high and significant relationship (p-value < 0.001). The AFDs of alpaca and llama fiber samples obtained with the two methods are very similar, because no significant difference was found at the t-test (p-value>0.172), and they have a strong, high and significant relationship between them, given the high Pearson correlation value (r ≥ 0.96 with p-value <0.001), high Concordance coefficient and bias correction factor. Similar results were found when PFT and OFDA100 were compared with S-Fiber Med. While in the results of our second objective, some visual differences were observed between the spectra of the different species. Accuracies up to 96.75% and 95.12% were obtained when evaluating the DNN and SVM models. Furthermore, an accuracy of 97.8% was obtained when evaluating the FTIR-ATR spectra of South American Camelids (SAC) fibers with DNN, and 97.2% when evaluating them with SVM. A 100% accuracy was obtained when evaluating the FTIR-ATR spectra of vicuña fibers with both models. No significant differences were found (p-value=0.368) by comparing the number of hits against the total number of alpaca, llama, vicuña, mohair and cashmere fibers using DNN. As per the results, it seems that DNN is more accurate than SVM. In conclusion, a novel system was developed that provides rapid (40 seconds/sample), precise, accurate, and automatic measurements of medullation and AFD. Additionally, FTIR-ATR spectrometry techniques combined with machine learning models are a reliable alternative for the identification of SAC and goats through the spectrum of their fibers.

The quality of alpaca textile fibre has great potential, especially if objectionable fibres (coarse and medullated fibres) that cause itching are reduced, considering that objectionable fibres can be identified by diameter and medullation types. The objective of this study was to estimate genetic parameters for medullar types and their respective diameters to evaluate the possibility of incorporating them as selection criteria in alpaca breeding programmes. The research used 3149 alpaca fibre samples collected from 2020 to 2022, from a population of 1626 Huacaya type alpacas. The heritability and correlations of the percentages of non‐medullated (NM), fragmented medulle (FM), uncontinuous medullated (UM), continuous medullated (CM), and strongly medullated (SM) fibres were analysed, also the fibre diameter (FD) for each of the medullation types. The heritability estimated for medullation types were 0.25 ± 0.01, 0.18 ± 0.01, 0.10 ± 0.01, 0.20 ± 0.01 and 0.11 ± 0.01 for NM, FM, UM, CM and SM, respectively. The genetic correlations for medullation categories ranged from 0.15 ± 0.03 to 0.66 ± 0.02 (in absolute values). The heritabilility estimated for fibre diameter (FD) of each of the medullation types were 0.29 ± 0.03, 0.27 ± 0.02, 0.35 ± 0.02, 0.30 ± 0.02, 0.25 ± 0.02 and 0.10 ± 0.02 for FD, FD_NM, FD_FM, FD_UM, FD_CM and FD_SM, respectively. The genetic correlations for fibre diameter of the medullation types ranged from 0.04 ± 0.04 to 0.97 ± 0.01. FD, NM and FM are the main traits to be used as selection criteria under a genetic index, since they would reduce fibre diameter, and also increase NM and FM, and, in addition reducing indirectly CM, SM, and SM_FD. Therefore, the quality of alpaca fibre could be improved.

Mejoras en el peso de vellón de sucio (PVe) y diámetro medio de la fibra (DMF) favorece los ingresos de los productores de alpacas; sin embargo, estas variables están relacionadas directamente, complicando el mejoramiento simultáneo de ambas características. Por tanto, se requiere buscar nuevos criterios de selección para la mejora simultánea de ambas características. El propósito del estudio fue caracterizar la densidad de fibras (DenFib), densidad de conductos pilosos (DenCond) y la relación N.° de fibras/conductos (R_Fib/Cond) como posibles criterios de selección. Asimismo, determinar el tipo de relación entre ellas y con el PVe y DMF, y construir y evaluar índices de selección formales bajo escenarios con diferentes mediciones disponibles, considerando en todos los casos PVe y DMF debidamente ponderados por sus pesos económicos, como objetivos de selección. Para ello, se tomaron muestras de fibras, imágenes de piel y se registraron el PVe de alpacas de cuatro rebaños criadas en Huancavelica y Caylloma-Arequipa (Perú) entre 2018 y 2020. Las características de densidad fueron obtenidas mediante el procedimiento denominado Fiber Den, mientras que el DMF se evaluó mediante el FIBER EC, y el PVe se registró con una balanza. En la construcción de los índices de selección para PVe y DMF se aplicó la teoría de índices de selección de Hazel, considerando los valores indicados en la bibliografía y en el presente trabajo. La DenFib, DenCon y R_Fib/Cond varía entre 8.8 a 60 fib/mm2, 3.8 a 21.2 cond/mm2 y 2.0 a 4.3 fib/cond, respectivamente. Se encontraron conductos pilosos de hasta 8 fibras, siendo los más frecuentes los conductos con 1, 2 o 3 fibras. Se concluye que la DenFib y DenCond tienen relación lineal y directa con el PVe e inversa con el DMF, y que la inclusión de la DenFib y DenCond como criterios de selección dentro de un índice de selección mejorarían la exactitud de 0.53 a 0.63, pudiéndose obtener mayores progresos genéticos entre 15 y 18%, con mejores beneficios económicos, que cuando solo se utiliza el PVe y DMF.