Twelve Labs updates Pegasus 1.5 for time-based video metadata
Twelve Labs introduced Pegasus 1.5, an update to its AI model that transforms video into structured, time-based metadata. This version utilizes schema-driven segmentation, custom evaluation metrics, and reinforcement learning to align with real-world video workflows.
Key Takeaways
- Pegasus 1.5 shifts from clip-based QA to time-based metadata generation.
- The model uses schema-driven segmentation to structure video output.
- Twelve Labs built custom evaluation metrics for Pegasus 1.5.
- Reinforcement learning is part of the training approach for workflow alignment.
Why It Matters
Pegasus 1.5 moves Twelve Labs from clip-based QA toward structured, time-based metadata, which is a more operational format for video applications. That matters because schema-driven segmentation and custom evaluation metrics suggest the model is being tuned to real workflow outputs, not just generic video understanding. For the broader streaming video stack, this points to AI tools that organize video into usable metadata rather than only answering questions about clips. The key signal to watch is whether Twelve Labs shows concrete examples of the schemas or evaluation metrics it used in Pegasus 1.5.
Read full article at twelvelabs.io
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source