Content authorization is a widely employed practice in health education categorization to prevent children from encountering inappropriate material. The conventional approach involves encrypting sensitive areas as nonsensical content, allowing decryption only when users possess authorized keys. Nevertheless, children should likely access these sensitive areas if their parents forget to log out of the system. To address this limitation, the proposed scheme introduces a real-time access control mechanism that employs online face recognition to determine whether users can access the content. Furthermore, the sensitive regions are encrypted using mask algorithm and decrypted by extracting facial features, reducing the maintenance cost associated with key management. Experimental results affirm the viability of the proposed scheme.