The idea of safety and security is producing a difficult task. The government, people, and stakeholder must need a reliable environment with good protection scheme. Hence, real-time object detection is very important to the society that lives in the urban areas. Many researchers have been found in the literature but more efficient and accurate framework is required. This research gives a real-time technique for object recognition in video surveillance system. The technique is generated using local binary patterns (LBP) and support vector machine (SVM) using MOT15 dataset. The introduced technique has been implemented using Python. The experimental outcome based on different performance criteria showed the better performance of introduced technique in terms of precision and accuracy on MOT15 dataset compared to existing techniques.