In this paper we present a real-time object detection in the warehouse, to predict collision arising from moving forklifts andraising alarm when necessary. There are many research using YOLO for real-time object detection like detecting person or cars for Advanced Driver Assistance System(ADAS). There are many cargoes in the warehouse. Employees need to collect them and deliver to other place. When the employees are driving the forklift with many cargoes or big shipments that may block there vision to see objects in front of them. The employee driving the forklift might not see cargoes stored at the corner while turning, causing accident. The system will identify objects in real-time, received through surveillance camera set at a height from where it can clearly capture required frames to predict collision. If the camera predict that there will be an imminent collision, it will sound the alarm.