Now a days to identify objects in video is a challenging task due to the real-time environment challenges. These challenges are illumination variation, cluttered background, rain, fog, dust, mist, etc. Such works are very helpful for security surveillance, transportation, agriculture, manufacturing, medical, healthcare, robotics, and multimedia applications. The BGS approach is an effective way of separating the foreground region from the background scene. It's often used in applications that operate in real-time for surveillance systems, motion detection, and recognition, or gestures-based pixel classification BGS that works by removing the static or dynamic background from the current frame to detect the foreground moving objects. So. This work focuses on video datasets to detect objects and remove the background scene by applying a background-subtraction approach. This work also focuses on the lack of a detailed description of the considered dataset. Here, a deep learning technique and BGS method are applied for object detection. This work enhances the existing GMM method and also compares it with advance GMM. The proposed advance GMM based method provides the better outcome as compared with existing method.