The development of artificial intelligence-based systems, particularly the field of computer vision more specifically the detection of objects always involves a process of training and testing. The quality of the object detection greatly influenced the training process object detection. Viola-Jones method is often used in object detection process where this method requires a classifier file obtained from the training process. This research is used to determine the trend of the number of the training process on the quality of bottle caps object detection fanta, sprite, Frestea, coca-cola and soda. Tendency obtained can be linear or exponential graph. the result of the research detection a bottle cap, happened saturation amount of test data so that a larger number no longer affect the quality of detection the bottle caps and the relationship between the number of files of positive file and negative file affect the accuracy in detecting an object.
Keyword: training, testing, artificial intelligence, Viola-Jones