Intelligent surveillance cameras with artificial intelligence (AI)-based video analytic function have become pervasive in recent years. They hold the promise of bringing high fidelity, contextually rich sensing into our home town and workplaces as a means of making our life smarter and safer. Despite remarkable and indisputable advances, AI cameras are still limited in the proper convolutional neural networks (CNNs) model, and more importantly, do not easily design a robust system architecture between edge device and (cloud) server for real-world applications. Towards addressing these limitations, we have developed an commercialized AI camera can be installed at any position that we use to recognize the license plate incorporating front-end and back-end intelligence. For intelligent front-end system, we designed three unique CNNs models on AI cmear for detecting license plate with its corner points and recognizing the characters and numbers sequentially. To increase the accuracy of AI functions on camera continuously, it is connect to back-end intelligence server where the current models on AI camera is updating with new incoming data in a continual process of adaptation. We conducted a series of experiments, showing high accuracy and versatility of the new architecture, while yielding robust results that can be practically implemented.