The masters thesis revolves around the development of Bayesian Network for classifying waste textiles, by taking off-set in predicting whether a textile is denim. The Bayesian Network uses data from three sensors: color camera, NIR sensor, and x-ray sensor. The data output of each sensor is investigated to evaluate feature extraction methods. Five features are chosen to constitute the input of the Bayesian Network: small and large metal buttons, short metal zippers, color and mass. The correlations between each feature and a textile being denim are analyzed through a 76 piece sample set. The network structure is defined based on the found correlations, and using a bayesian dirichlet equivalent uniform prior, the Bayesian Network correctly classifies in average 80\% of the textiles in the sample set. Uncertainties of feature extraction are introduced to the sample set data, resulting in similar performance. Then, the model is compared to a sorting tree logic, which would perform worse than the Bayesian Network, leading to the conclusion that the Bayesian Network shows promise in classification of waste textiles. Lastly, the methodology is generalized to be applied in other classification use cases.