The automatic sorting system has been reported to be complex and a global problem. This is due to the sorting machines’ failure to include flexibility in their design concept. Furthermore, products or objects may have defects and be in poor shape when they reach the hands of customers. For instance, if products on a conveyor belt are not in perfect posture or position, the robot’s arm might be unable to grasp them effectively, resulting in unanticipated incidents such as objects falling or deformities due to a wrong grabbing technique. The advancement of computer vision technology is making it easier to solve this challenge with accurate results. A machine vision solution using the conventional Canny edge detector technique can be a good way to detect object deviation based on the object’s edges. The algorithm is improved by incorporating the deviation angle detection technique to determine whether the objects are in the proper position for the robotic arms to grab. This research commits to developing an object deviation detection algorithm that can inspect the posture of things on the conveyor and provide information about the deviation angle. This can ensure the system’s smoothness and efficacy when employed as a sorting system in industry. To evaluate this algorithm’s effectiveness when applied in real time, it will be deployed to an embedded device. The NVIDIA Jetson Nano, a GPU-based device, has been selected as the target hardware. The edges and orientation angles of objects have been precisely detected by the suggested system.