The purpose of this research is to build a system that can detect certain face among several people. The system works using face recognition in searching suspects from a database. 30 images and 9 videos are used for training and testing. In the training process carried out several stages: (1) pre-processing of face images, (2) extraction of facial image training features using the Histogram of Oriented Gradient (HOG) method, (3) face classification using the SVM method. There are several steps for testing process i.e.: (1) video frame extraction, (2) cascade classifier method with the Local Binary Pattern (LBP) feature descriptor, (3) HOG for extracting facial features detected in the frames, and (4) SVM as the face classifier. The processed video data is 1920x1080 pixel resolution which has been recorded using a CCTV camera which is mounted on a wall with height and slope angle of 2.5 meters and 60 degrees. There are two goals implemented in this research. The first goal is to find the highest accuracy from several testing of frame sampling by setting the best threshold value. The second goal is to find the lowest processing time.