This release introduces S3-based frame storage, deployment hardening, and documentation improvements.
- RTP Timestamp Alignment — Fusion analytics now uses RTP SRC time field to match frames with the nearest metadata records for improved synchronization.
- SeaweedFS S3 Integration — DL Streamer now stores output frames and images in an S3-compatible SeaweedFS backend, with full Helm chart support.
- Vision Metadata Persistence — DL pipeline vision metadata is now persisted to InfluxDB via fusion analytics for improved traceability.
- Helm Deployment — Helm charts for multimodal deployment are now available.
- Simplified weld-data-simulator deployment by embedding simulation data directly into the container image, removing the external PV/PVC volume dependency.
- Updated system requirements to reflect CPU-only validated configurations.
- Updated third-party service images: Telegraf, Grafana, Eclipse Mosquitto, MediaMTX, Coturn, and SeaweedFS.
- Hardened SeaweedFS container runtime
- Added S3 storage access and credential setup guides.
- Reorganized how-to guides and navigation/toctree structure.
- Updated Helm deployment documentation.
- Fixed duplicate headings and TOC build issues.