Nowadays, the rapidly changing of manufacturing environment has pushed companies to achieve more customer satisfaction by enhancing product quality, reducing production cost, and realizing sustainability. Anomaly detection has a strong influence on the quality of products and it is usually conducted through visual quality inspection. The visual quality inspection of a product can be performed either manually or automatically. The manual inspection suffers from being a monotonous task, leading to overlooked errors and subjective assessments. Accordingly, the manufacturing industry has high ambitions to rely upon automated quality inspection systems to cope with the requirements of smart manufacturing and the emergence of industry 4.0. Efficient utilization of big data can enable the development of intelligent quality inspection systems. Machine learning as one of the prevailing data analytics methods is widely used to support and improve the performance of the automated quality inspection systems. This research compares the performance of Recurrent Neural Networks (RNN) like Multilayer Perceptron (MLP) with the traditional machine learning algorithms (TMLA) for anomalies detection in manufacturing such as Decision Trees, Random Forest (RF), k-Nearest-Neighbor (KNN), Support Vector Machine (SVM), Naïve Bayes, and Logistic Regression (LR). A data set for faults is adapted from the literature to fairly compare the performance of these algorithms considering different accuracy measures such as accuracy, precision, sensitivity, and F1-score.