Indonesia's ceramic tiles production is currently ranked between fifth and sixth of the world, that showed ceramic tiles is one of the largest commodities in Indonesia. So, ceramic quality becomes very important to be considered in it used as one of the basic building materials. The existing ceramic testing system in Balai Besar Keramik is performed by the operator repeatedly. With the repetitive perform, it causing fatigue to operator that resulting in decreased work ability, so in calibration and readings of measuring instruments occur an error. To improve accuracy level of quality control system to remain stable, it’s necessary an intelligent device that can overcome errors that occur in the inspection process using human vision on surfaces defect detection of ceramic tiles that have a high level of system stability, therefore required automation system design of image processing-based using Artificial Neural Network method with backpropagation algorithm. Data that need for processing system, using 45 data that divided into data training and data testing with the proportion of ratio 70% and 30%. Process of system design use two modes, namely offline and realtime mode that obtained accuracy rate of 96.9% for offline mode and 92.3% for realtime mode.