Protective coatings are essential to protect structures by isolating the underlying materials, or substrate, from harmful environmental factors. When coatings degrade, the exposed substrate deteriorates, which leads to costly maintenance and potential safety risks. Current paint condition assessment techniques rely heavily on visual inspection or using non-destructive techniques like eddy current and ultrasonic testing. These methods, although effective, are not practical when inspecting large surfaces. In this study, a comprehensive framework that combines hyperspectral imaging (HSI) and machine learning (ML) was developed to assess the condition and predict the thickness of paint coating under accelerated aging tests. Aluminum plates painted with aliphatic polyurethane-based paint have been prepared with different paint thicknesses and various aging states. The HYSPEX SWIR 384 camera was used for data collection within a spectral range of 930-2505 nm. The QUV, a UV weathering tester manufactured by Q-Lab, was used to age the samples according to ISO 11507 standard with a maximum period of 1000 hours. Various ML and deep neural network (DNN) models were explored to predict paint thickness. The DNN model, with four hidden layers, resulted in an RMSE of 11.6 µm and R2 of 0.91, predicting thicknesses from 43 to 218 µm. For the degradation assessment, the DNN model delivered the highest performance with an R2 of 0.896 and RMSE of 160 hours in predicting aging hours. The obtained results demonstrated that the developed models capture well the relationship between paint thickness/degradation and the reflectance values, highlighting the effectiveness of HSI for paint condition assessment.