License Plate Recognition plays a pivotal role in modern traffic law enforcement, ensuring public safety and order. However, conventional surveillance systems such as CCTVs and static cameras lack real-time response capabilities and have limited mobility. With the growing number of vehicles, the requirement for automated and mobile methods for license plate recognition has soared. The noisy and dynamic environment of license plates further exacerbates this issue. While Deep Learning (DL) can help automate such tasks, the computational demands of DL pose a significant hurdle for real-time usage and mobility. In this regard, the declining costs and enhanced computational capabilities of microcontrollers offer promising potential for enabling the implementation of DL-based techniques in license plate recognition in resource-constrained scenarios. This paper introduces an approach for automated license plate recognition designed to guarantee mobility and real-time responsiveness. The proposed framework integrates various elements, encompassing microcontrollers, Internet of Things (IoT), Deep Neural Networks, and computer vision technologies. Furthermore, to alleviate the computational overhead on the microcontroller, the system leverages Transfer Learning and Cloud Computing for enhanced efficiency. The system was tested for real-time performance using a camera onboard a microcontroller, which was used to detect the license plates. The system delivered good accuracy for license plate recognition, both across existing datasets and for real-time images on multiple metrics. This system can also be integrated with wearable devices such as helmets or goggles and used by traffic law officials to facilitate easy monitoring and surveillance of traffic laws.