WiMi Hologram Cloud
WiMi Hologram Cloud Develops Intelligent Face Detection and Database Matching Recognition System
WiMi Hologram Cloud a leading global Hologram Augmented Reality (“AR”) Technology provider, announced the development of a holographic intelligent face detection and database matching recognition system, which identifies and authenticates people through the distribution of facial feature points in an image. Through image processing technology, features related to personal characteristics are extracted and analyzed, converted into digital form, and processed by computer for the identification or personal verification.
The system uses a camera to collect facial information. Pre-processed by filtering and RGB/RGBD technology, transcoding, 3D alignment and reconstruction, face detection, area segmentation, and face tracking are completed before feature extraction and matching. To improve accuracy and recognition efficiency, WiMi’s R&D team independently developed face detection algorithm that achieves real-time detection of facial features and reduces false detection using real-time dynamic data enhancement technology. This approach has significantly improved the speed and accuracy of face recognition matching. WiMi’s system can be applied to a variety of complex scenarios. WiMi has developed technologies for face quality assessment, pose assessment, and backlight detection based on its characteristics in complex environments, and has proposed a new method based on face-first push mapping.
Regarding face attributes, the system uses a multiple attribute association approach, where details with high correlation are used for sharing network features, while mutual exclusivity is used for feature decoupling. The algorithm can predict individual models and multiple facial features online and in real time through repeated iterations. The system uses a deep learning-based non-matching dynamic scene recognition algorithm, which achieves accurate recognition by simplifying the network and combining it with self-developed facial pose and face correction technology, which allows facial features in complex scenes to be filtered and reconstructed to solve the problem of face occlusion.
With the continuous development of artificial intelligence and big data, face information is becoming an increasingly important part of personal information. WiMi’s system ensures user information security through data desensitization, privacy calculation, and decentralized storage. With full authorization from the user, the user’s facial features are identified, tagged, and encrypted using non-reversible cryptography, thus ensuring the security of data transmission and storage and adequate protection of the user’s facial features.
Using privacy computing technology such as Trusted Execution Environment and Secure Multiparty Computing, the secure processing of facial features is achieved without collecting and sharing the original data, and only external desensitization is performed. It ensures that there will not be any leakage in the use, processing, and circulation, effectively solving the contradiction between data confidentiality and efficient processing and circulation. Secure isolation and decentralized storage enable differential privacy, ensuring that attackers cannot infer other information from specific data.
