Abstract :
The training process requires very high computer resources and time
which is directly proportional to the amount of training data. The amount of
training data determines the accuracy of object detection. The long training
process is caused because the computer has low specifications and the distribution
of training files will speed up the process of vector file formation, minimize errors
when cutting Haar features on positive objects and also minimize errors that occur
during the training process. The problems that arise next are how to overcome this
so that a better knowledge is obtained. This study provides suggestions and
analysis results of the process of merging knowledge acquisition and its effect on
the accuracy of object detection using the Viola-Jones method with the final result
an error rate of 23,689% and an object detected rate of 65,584%. Viola Jones
method uses the file classifier obtained from the training process