The Post-harvest coffee bean selection plays a crucial role in ensuring optimal bean quality during production processes. Currently, this process is manually conducted in Indonesia. Implementing computer vision can enhance the objectivity of the sorting process through machine vision. An effective object detection system is essential to support a prototype coffee bean quality classification system based on NVIDIA Jetson Nano. The Hue Saturation Value (HSV) color filter method proves effective in detecting objects within a given image frame. Performance evaluation is conducted by assessing the alignment between workflow design and system operation. While the webcam-based object detection system successfully deployed, its effectively identifies coffee bean objects, it faces limitations in detecting smaller, dark-colored beans beyond the specified HSV color threshold. These limitations are attributed to the webcam's specifications, including its rolling shutter, which results in a 'jello effect' when dealing with moving objects.