In many electronics manufacturing processes, quality control of the cable assembly is conducted manually by a human operator. However, given that the sizes of cables are thin and the different product groups include complex color combinations, determining whether the correctly colored wire is assembled into the right port is challenging for a human operator. With recent developments, automated visual quality control systems in the manufacturing industry can accurately detect errors and lower costs. These systems use deep learning models, which usually demonstrate higher prediction performance when trained by large datasets of images collected under various working conditions ranging from simple to complex. As preparing extensive datasets entails high costs and time, open-source datasets related to the subject with real-life images are helpful to enrich the image dataset. This study uses an industrial dataset and an open-source repository to investigate the prediction performance of You Only Look Once version 8 with Segmentation (YOLOv8Seg) object detection model to contribute to quality control processes in cable assembly in the electronic manufacturing sector. The preliminary findings of the study demonstrate that YOLOv8Seg shows superior performance in detecting different colored wires in the electronics assembly with a mean average precision above 92%.