Explore the Future of Generative AI with YUAN at Smart City Expo World Congress

At Smart City Expo World Congress in Barcelona, video technology is being demonstrated as a core driver of smart city development. Vision AI and visual AI agents not only enhance urban safety but also signiﬁcantly boosts operational eﬃciency. In this exhibition, YUAN, as a leader in video technology, showcases its latest advancements in the integration of video capture and AI, painting an ideal blueprint for the future of smart cities.

Challenges and Breakthroughs in Diverse Video Sources

Video applications in smart cities have long surpassed the basics of traditional IP cameras and NVR recording. With advancements in technology, cities now require a more diversiﬁed range of real-time video sources. YUAN points out that although next-generation network cameras have become widespread, cost-eﬀective options like TVI, SDI, and CVBS cameras remain the primary choice for system integrators. Furthermore, MIPI and GMSL cameras in the unmanned aerial vehicle (UAV) market are gradually becoming mainstream in emerging applications. While such diverse video sources enrich data support for cities, they also increase the complexity of video integration, posing new challenges for the fusion of AI applications.

Hardware Innovation: The Power of NVIDIA Accelerated Computing at the Edge

YUAN's solutions, powered by the NVIDIA Jetson Orin platform for edge AI and robotics, showcased at the exhibition, elevate smart city video processing to new heights. The Air series integrates the NVIDIA Jetson Orin NX module onto a business-card-sized board, featuring 12 diﬀerent video capture modules that support IP, TVI, SDI, GMSL, HDMI, and more. Customers can ﬂexibly swap video input modules based on on-site needs, enabling real-time, eﬃcient video processing. This innovation earned the Air series the 2024 Computex Golden Best Award. At this Smart City Exhibition, YUAN is proud to share its best-in-class design.

Additionally, YUAN's Near, Edge, and Air product lines oﬀer comprehensive edge computing solutions, ranging from 20 TOPS to 1700 TOPS, with options spanning the NVIDIA Jetson and NVIDIA IGX platforms to meet diverse computational needs across application scenarios.

AI Innovation: High-Resolution, Low-Latency NVIDIA Jetson Orin Edge Computing-Based Camera

To meet the demand for high-performance video equipment in smart cities, YUAN has collaborated with Sony to launch a 4K60 camera with 25x optical zoom, integrated with the NVIDIA Jetson Orin NX module. This camera, equipped with 4K60 resolution, HDR, and IP66 waterprooﬁng, is durable and reliable in outdoor environments. Additionally, built-in RDMA technology and NVIDIA CUDA-powered video processing capabilities eﬀectively reduce latency. The Jetson Orin NX-equipped camera enables software developers to deploy AI analysis directly on the 100 TOPS camera end, enhancing real-time processing eﬃciency and expanding possible application scenarios.

Generative AI: Transforming a New Era of Smart NVR and Camera Applications

Generative AI is redeﬁning the way smart NVRs and cameras are utilized. NVIDIA Jetson Orin brings multimodal generative AI capabilities to smart cities. Supported by visual language models (VLMs), the platform integrates diverse inputs such as voice, text, and video, providing accurate scene descriptions and behavior recognition for smart cities. At this exhibition, YUAN's SmartVDO NVR software demonstrates the application of generative AI in DVR and VMS systems, oﬀering forward-looking solutions for smart security.

Intelligent Understanding: Integration of VILA/CLIP Technology with NVR

YUAN's AI team has integrated VLM VILA technology with NVIDIA Jetson AGX Orin's network video recorder software using NVIDIA Metropolis for vision AI capabilities, allowing the NVR system to interpret real-time video content, including for video captioning in hazardous safety situations. Additionally, YUAN has applied the CLIP backend's model for video vectorization, storing the data in VectorDB. This allows video managers to perform instant searches based on speciﬁc criteria, revolutionizing traditional perceptions of NVR capabilities.

Smart Interaction: Breakthrough Applications of Whisper/Llama

Beyond video recognition, YUAN further integrates Whisper voice technology, enabling NVR systems with voice interaction capabilities, eliminating the need for keyboard operation, and making NVR systems more user-friendly. Moreover, Llama adds intelligent semantic search functionality, replacing the traditional keyword search model, making video management interactions more eﬃcient and convenient.

Conclusion: Opening a New Chapter in Smart City Video Technology

By combining generative AI technology with professional video expertise and hardware design, YUAN brings a revolutionary model to smart city video processing. These innovative solutions not only address the challenges of diverse video sources but also oﬀer eﬃcient experiences in real-time analysis, voice interaction, and intelligent search. With the support of the NVIDIA Jetson Orin platform, YUAN is redeﬁning video applications in smart cities, providing smarter and more eﬃcient video solutions for urban development.