2022 Bilibili Technology Patent Selection Awards

This article introduces the 2022 Bilibili Technology Patent Selection Awards, held to celebrate World Intellectual Property Day and promote intellectual property awareness within the company. The event featured two categories: Best Popularity Award and Most Popular Award, with five patent projects selected in each category.

Best Popularity Award:

1. Advanced Bullet Comments : This project involves various display forms of advanced bullet comments, including parallel bullet comments (dividing comments into two camps for parallel reverse movement) and four-way bullet comments (dividing comments into four camps for movement in four directions from the screen center).

2. Gift Animation Display : This scheme optimizes gift animation display by pre-dividing the live room into multiple playback areas, allowing gift animations to play only in corresponding areas to control device performance overhead.

3. Video Rendering : This project optimizes video rendering effects by establishing a communication mechanism between the engine scene layer and player to efficiently obtain video playback data for graphic processing operations.

4. Magic Mirror Cloud Testing Platform : This comprehensive testing platform includes device scheduling (using cloud platform tagging for terminal device scheduling), model detection (automatically identifying incompatible terminal models through image similarity calculation), pop-up handling (coordinate-based pop-up processing), and overdraw detection (automatically testing page images for overdraw using color detection).

5. Virtual Avatar Video Production : This scheme optimizes video production by importing virtual avatars into video editors, editing them based on user operations and audio information, and synthesizing them with audio to create virtual avatar videos.

Most Popular Award:

1. Virtual Material Editing : This project optimizes virtual material editing by allowing users to customize template decoration parts and automatically applying them to corresponding virtual avatar parts based on category identification.

2. Mini Program Features : This comprehensive scheme includes business reuse (combining H5 containers with client programs), channel compatibility (adapting different channel page differences), and mini-game rendering (binding H5 game data rendering with mobile OpenGL ES rendering for performance optimization).

3. Cross-Live Streaming Switching : This project establishes a complete switching mechanism for live streaming scenarios, including AI-based automatic switching based on video stream health, resolution/frame rate control, and frame judgment to avoid re-encoding.

4. Blur Image Face Enhancement : This scheme enhances facial areas in blurry images using transition zones, enhancement masks, and image segmentation networks to reduce negative effects from non-face area enhancement.

5. Ghost Video Production Tools : This comprehensive project includes AI speech recognition for video material processing, video slicing with Bezier curve speed variation, automatic pitch detection and adjustment, and specialized editors for post-production correction.

The article concludes by inviting readers to choose their favorite project and expressing Bilibili's commitment to fostering technical innovation and atmosphere within the company.