deepstream-sop - DeepStream SOP compliance inference service
Build, deploy, evaluate, debug, and measure latency for a GPU-accelerated service that checks industrial video against SOP step sequences.
Tags
Updated: 2026-10-06Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Build DeepStream SOP pipelines
- Deploy the inference microservice
- Evaluate procedure compliance
- Debug pipeline components
- Measure streaming latency
- Detect event boundaries
- Classify video chunks
- Validate SOP step sequences
- Stream inference results
- Publish Kafka messages
- Expose Prometheus metrics
- Process industrial camera feeds
Inputs
- Video files
- RTSP streams
- Basler camera feeds
- SOP step sequences
- Model configuration
- Environment variables
- Local reference code
Outputs
- SSE streaming responses
- Non-streaming chat responses
- Kafka JSON messages
- Kafka NvProto messages
- Prometheus metrics
- Optional RTSP output
- SOP compliance results
Requirements
- GPU-enabled DeepStream environment
- Docker runtime
- Triton support
- vLLM model support
- FastAPI runtime
- Kafka service
- Basler Pylon SDK for cameras