Accuver's AI Tool Provides Repeatable MOS Without A Reference Video
Accuver has published a case study on its XCAL-VQML product, which was deployed at a major network solution vendor's R&D center. The AI-driven solution provides no-reference video quality assessment, generating real-time Mean Opinion Score (MOS) values to measure user experience. The deployment reportedly produced reliable and repeatable MOS values for streaming video without requiring a reference file, allowing the customer to benchmark video services.
Key Takeaways
- The customer's challenge was establishing a consistent testing environment to measure how network performance affects perceived video quality.
- Accuver's XCAL-VQML uses an AI model, trained on large video datasets, to predict MOS values in real-time without a reference source.
- The system uses a 'Virtual Camera' capability, replacing physical camera input with standardized reference files to ensure objective testing.
- The deployment reportedly produced repeatable MOS values, giving the R&D team confidence to benchmark services and plan for wider adoption across its global labs.
Why It Matters
No-reference quality assessment aims to convert abstract network KPIs into a scalable, repeatable measure of perceived user experience. This allows engineering teams to quantify the impact of network changes on video quality in a way that traditional, manual, or camera-based MOS testing cannot. For operators and equipment vendors, this creates a consistent methodology for benchmarking services against competitors, especially in live environments where a pristine reference source is unavailable. What to watch is whether these AI-driven MOS metrics move from R&D labs into the live production monitoring dashboards used by major streaming services and network providers.
Read full article at accuver.com