The increasing integration of networked Electronic Control Units (ECUs) in modern vehicles has elevated the importance of securing in-vehicle communication protocols, particularly the Controller Area Network (CAN) bus, which remains vulnerable to a range of cyberattacks. This paper proposes a novel Payload Image-Based Intrusion Detection System (PIB-IDS) to address this challenge. The proposed method encodes CAN identifiers and payload data into compact image representations, which are subsequently processed by a Convolutional Neural Network (CNN) for the binary classification of CAN traffic as benign or malicious. To ensure applicability in resource-constrained embedded environments, the CNN model is optimized through quantization techniques, achieving a trade-off between computational efficiency and detection accuracy. Extensive experiments conducted on publicly available CAN datasets demonstrate the system’s robustness against diverse attack scenarios. Furthermore, real-time deployment on the STM32F746 microcontroller confirms the feasibility of the proposed PIB-IDS for practical automotive applications, validating both its hardware compatibility and operational efficiency.