The primary goal of inspection is to detect faulty products and prevent them from being delivered in the market. In the textile industry, manual inspection carried out by human inspectors is not only labor intensive but also prone to errors. Given the need to maintain production efficiency and ensure proper inspection, there is a demand for an automated approach to conduct inspection on textile products. This paper proposes a neural network-based binary classifier to detect presence of defect on yarn and knitted fabric, using texture features derived from Gray Level Co-occurrence Matrix and Local Binary Patterns. The final proposed model is deployed as a REST (Representational state transfer) API endpoint on a desktop application, where user can upload fabric images to obtain prediction. Using the proposed model, an accuracy rate of up to 83.9% and 70% were achieved on validation set (N=100) and test set (N=43) respectively, suggesting that the model can perform relatively well at capturing defect cases.