Image Understanding And Multimedia Content Generation
OVERVIEW
With the advancement of AI deep learning technology, image recognition has achieved significant breakthroughs, advancing into a new era of image cognition and understanding. Our institute focuses on key research areas like facial recognition and behavior analysis, enabling precise identification and behavior interpretation for applications in security monitoring, personnel management, and smart law enforcement.We are also committed to multimedia content generation, producing diverse outputs such as images, videos, animations, and sound effects. These innovations enhance image processing and creative content production, paving the way for smarter cities and a new era of digital living.
Image Understanding And Multimedia Content Generation
CORE TECHNOLOGY
- Face Recognition
- Human and Behavior Recognition
- Digital Human Generation
- Multimedia Content Generation
Face Recognition Applications
Human and Behavior Recognition Applications
Digital Human Generation Applications
Multimedia Content Generation Applications
Image Understanding And Multimedia Content Generation
APPLICATION STATUS
Face Recognition：Based on face recognition, We facilitates the development of innovative applications such as identity verification and person attributes (gender/age/ facial features). We even developed mask detection during the COVID-19 pandemic. Additionally, our team achieved first place in Taiwan on US NIST FRTE 1:1 evaluation held in October 2024, an international competition that focuses on improving accuracy of face recognition. For business, we provide solutions for company access control system, customer analysis, identity verification, no contact number ticket system and airport management. Furthermore, edge AI is currently one of the most trending technologies that is currently being used worldwide. We are now extending Face Recognition on all-in-one machine.
Human and Behavior Recognition：Powered by AI deep learning algorithms, human detection delivers human and behavior information, such as person appearance, person location, people number information and people interactive information. We develops smart electronic fence system which improves traditional electric fence efficiency. We also provide real-time smart electronic fence alert service. The application fields include critical infrastructure, MRT/railway station, scenic area, etc. Besides, we also plan to deploy human detection on embedded system. We hope develop lower-cost, lower-power and faster solutions in the future.
Digital Human Generation：This technology can integrate with voice recognition, knowledge-based Q&A, voice mimicry, and text-to-image generation to create interactive digital humans in real time. It features natural gesture interactions, precise lip-syncing for Chinese, and the ability to adapt gestures based on the length of audio files. It serves as a solution for promoting knowledge or products, enabling organizations or individuals to establish dedicated knowledge bases and craft personalized avatars. Through a Q&A format, it facilitates information dissemination and marketing communication.
Multimedia Content Generation：In multimedia generation, we leverage state-of-the-art AI models to create diverse content, including images, videos, animations, and sound effects. By training with localized data, we infuse unique Taiwanese elements into high-quality creative outputs. For image and video understanding, we focus on multimodal language models and advanced technologies like discriminative AI and Edge AI, enabling innovative applications such as human activity and behavior recognition. These innovations are applied in smart security monitoring and creative content production, highlighting Taiwan's technological expertise and localized capabilities.