Forklifts, commonly used in modern factories for transporting goods, are often driven with agility, which can lead to safety hazards. In this study, we present a real-time, on-board monitoring system based on edge computing and deep learning to assist forklift operators for more safe driving. Serving as the forklift’s peripheral vision, a wide-angle imaging module with three cameras is designed to detect pedestrians and designated object (e.g. traffic cone, oil drum, truck, bicycle, etc.) around the vehicle. This system actively identifies pedestrians and objects around the forklift’s path to warn the operator and enhance safety. Additionally, a wide-angle camera is also installed in front of the operator to monitor the driver’s driving behavior. The system processes images at a speed of over 20 frames per second, making it suitable for real-time pedestrian detection and collision avoidance.