Manufacturing industries are facing major challenges in terms of improving product quality and increasing throughput while sustaining production costs to acceptable levels. Product-oriented processes, both legacy and new, are poorly monitored and controlled on the basis of distributed loop controllers that are aiming to maintain critical process variables within acceptable bounds. Thus, poor quality product results when such processes are subjected to large disturbances - operational failures, environmental changes, and changes in loading conditions. In this research, product quality modeling and control based on a vision inspection methodology is proposed to improve product quality and increase productivity. The main contributions of this research are twofold. First, this research introduces a product quality modeling methodology that combines both physical-based modeling and data-driven modeling. The quality model is the link between information coming from the inspection of product features and the specification of process control strategies. It is essential to control and optimize the process. Physical-based modeling is used to model the product temperature profile, and data-driven modeling is used to train the mapping from the product temperature profile to each quality metric. The break down of the sub models increase the flexibility of model development and reduce the effort to change the model when the quality metrics change. The second contribution is the development of a novel approach to control product quality based on vision inspection, which is developed as part of a hybrid, hierarchical architecture. The high-level control module involves scheduling of multiple plant processes, diagnostics of the failure condition in the process, and the supervision of the whole process. The mid-level control module, which is the focus of the work presented here, takes advantage of baking product quality indicators and oven parameter measurements to optimize zone temperature and conveyor speed set points so that the best product quality is achieved even in the presence of disturbances. The low-level control module consists of basic control loops. Each of them controls parameters of each operation in the process separately. They are generally simple and easy to implement.