# madhavan-hos/sdv-intel-edge-pipeline architecture

> Architecture diagram of the GitHub repository madhavan-hos/sdv-intel-edge-pipeline, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-06). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/madhavan-hos/sdv-intel-edge-pipeline
- Repository: https://github.com/madhavan-hos/sdv-intel-edge-pipeline

## Overview

This is an SDV vision workshop stack: train or export a YOLO model on HPC, move an OpenVINO artifact to the edge, and combine RealSense obstacle-distance perception with FHD-camera lane detection in a local dashboard. A separate socket-based path supports remote HPC inference. The README documents model transfer and the major runtime choices; sampled code shows the sensor processing and remote request/response. Some setup and training details are represented only by documented capabilities, not inspected implementation.

Main entry points: [`session_09_integration/local_sdv_pipeline_optimized.py`](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py) (Responsive dashboard), [`models/best_openvino_model`](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/tree/main/models/best_openvino_model) (OpenVINO model), [`session_09_integration/edge_client_streamer.py`](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/edge_client_streamer.py) (Edge frame streamer), [`session_05_model_hpc/export_openvino.py`](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/export_openvino.py) (OpenVINO export)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_model["Model preparation"]
  node_yolo_training["YOLO training<br/>[train_yolo.py]"]
  node_standard_export["OpenVINO export<br/>[export_openvino.py]"]
  node_precision_export["Precision variants"]
  node_model_files["OpenVINO model"]
end

subgraph group_sensing["Edge sensing"]
  node_camera_discovery["Camera discovery<br/>[camera_scanner.py]"]
  node_rgb_depth["RGB-depth alignment"]
end

subgraph group_perception["Perception"]
  node_lane_usecase["Lane detector<br/>[lane_detection.py]"]
  node_obstacle_usecase["Obstacle detector"]
  node_device_benchmark["Device benchmark"]
end

subgraph group_runtime["Integration paths"]
  node_local_pipeline["Local dashboard"]
  node_responsive_pipeline["Responsive dashboard"]
  node_capture_worker["Camera workers"]
  node_inference_worker["Detection worker"]
  node_lane_worker["Lane worker"]
  node_hpc_server["HPC inference server"]
  node_edge_client["Edge frame streamer"]
end

node_hpc_user(("Workshop operator"))
node_edge_cameras(("Edge cameras"))
node_hpc_host(("HPC host"))

node_hpc_user -->|"trains model"| node_yolo_training
node_yolo_training -.->|"exports weights"| node_standard_export
node_standard_export -.->|"writes model"| node_model_files
node_hpc_user -->|"exports weights"| node_standard_export
node_hpc_user -->|"exports variants"| node_precision_export
node_hpc_user -->|"transfers model"| node_model_files
node_camera_discovery -->|"scans devices"| node_edge_cameras
node_edge_cameras -.->|"provides streams"| node_rgb_depth
node_edge_cameras -.->|"provides frames"| node_lane_usecase
node_edge_cameras -->|"provides RGB-depth"| node_obstacle_usecase
node_rgb_depth -->|"provides aligned depth"| node_obstacle_usecase
node_hpc_user -->|"benchmarks devices"| node_device_benchmark
node_hpc_user -->|"starts dashboard"| node_local_pipeline
node_hpc_user -->|"starts dashboard"| node_responsive_pipeline
node_model_files -->|"loads model"| node_local_pipeline
node_model_files -->|"loads model"| node_responsive_pipeline
node_responsive_pipeline -->|"starts workers"| node_capture_worker
node_responsive_pipeline -->|"starts workers"| node_inference_worker
node_responsive_pipeline -->|"starts workers"| node_lane_worker
node_capture_worker -->|"publishes frames"| node_inference_worker
node_capture_worker -.->|"provides camera frames"| node_lane_worker
node_hpc_user -->|"starts remote mode"| node_edge_client
node_edge_cameras -->|"provides frames"| node_edge_client
node_edge_client -->|"sends frames"| node_hpc_server
node_hpc_server -->|"returns annotated frames"| node_edge_client
node_hpc_host -.->|"hosts inference"| node_hpc_server

click node_yolo_training "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/train_yolo.py"
click node_standard_export "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/export_openvino.py"
click node_precision_export "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/optimization/export_openvino_precisions.py"
click node_model_files "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/tree/main/models/best_openvino_model"
click node_camera_discovery "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_04_vision/camera_scanner.py"
click node_rgb_depth "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_06_fusion/rgb_depth_fusion.py"
click node_lane_usecase "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_08_usecase/lane_detection.py"
click node_obstacle_usecase "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_08_usecase/obstacle_detection.py"
click node_device_benchmark "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_07_acceleration/benchmark_devices.py"
click node_local_pipeline "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline.py"
click node_responsive_pipeline "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py"
click node_capture_worker "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py"
click node_inference_worker "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py"
click node_lane_worker "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py"
click node_hpc_server "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/hpc_server_inference.py"
click node_edge_client "https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/edge_client_streamer.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_yolo_training,node_standard_export,node_precision_export,node_model_files toneBlue
class node_camera_discovery,node_rgb_depth toneAmber
class node_lane_usecase,node_obstacle_usecase,node_device_benchmark toneMint
class node_local_pipeline,node_responsive_pipeline,node_capture_worker,node_inference_worker,node_lane_worker,node_hpc_server,node_edge_client toneRose
class node_hpc_user,node_edge_cameras,node_hpc_host toneIndigo
```

## Components

### Model preparation

- [YOLO training](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/train_yolo.py): `session_05_model_hpc/train_yolo.py`
- [OpenVINO export](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/export_openvino.py): `session_05_model_hpc/export_openvino.py`
- [Precision variants](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_05_model_hpc/optimization/export_openvino_precisions.py): `session_05_model_hpc/optimization/export_openvino_precisions.py`
- [OpenVINO model](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/tree/main/models/best_openvino_model): `models/best_openvino_model`

### Edge sensing

- [Camera discovery](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_04_vision/camera_scanner.py): `session_04_vision/camera_scanner.py`
- [RGB-depth alignment](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_06_fusion/rgb_depth_fusion.py): `session_06_fusion/rgb_depth_fusion.py`

### Perception

- [Lane detector](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_08_usecase/lane_detection.py): `session_08_usecase/lane_detection.py`
- [Obstacle detector](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_08_usecase/obstacle_detection.py): `session_08_usecase/obstacle_detection.py`
- [Device benchmark](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_07_acceleration/benchmark_devices.py): `session_07_acceleration/benchmark_devices.py`

### Integration paths

- [Local dashboard](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline.py): `session_09_integration/local_sdv_pipeline.py`
- [Responsive dashboard](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py): `session_09_integration/local_sdv_pipeline_optimized.py`
- [Camera workers](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py): `session_09_integration/local_sdv_pipeline_optimized.py`
- [Detection worker](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py): `session_09_integration/local_sdv_pipeline_optimized.py`
- [Lane worker](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/local_sdv_pipeline_optimized.py): `session_09_integration/local_sdv_pipeline_optimized.py`
- [HPC inference server](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/hpc_server_inference.py): `session_09_integration/hpc_server_inference.py`
- [Edge frame streamer](https://github.com/madhavan-hos/sdv-intel-edge-pipeline/blob/main/session_09_integration/edge_client_streamer.py): `session_09_integration/edge_client_streamer.py`

### Other components

- **Workshop operator**
- **Edge cameras**
- **HPC host**

## Connections

- Workshop operator → YOLO training: trains model (evidence: `README.md`)
- YOLO training → OpenVINO export: exports weights
- OpenVINO export → OpenVINO model: writes model
- Workshop operator → OpenVINO export: exports weights (evidence: `README.md`)
- Workshop operator → Precision variants: exports variants (evidence: `README.md`)
- Workshop operator → OpenVINO model: transfers model (evidence: `README.md`)
- Camera discovery → Edge cameras: scans devices (evidence: `session_04_vision/camera_scanner.py`)
- Edge cameras → RGB-depth alignment: provides streams
- Edge cameras → Lane detector: provides frames
- Edge cameras → Obstacle detector: provides RGB-depth (evidence: `session_08_usecase/obstacle_detection.py`)
- RGB-depth alignment → Obstacle detector: provides aligned depth (evidence: `session_06_fusion/rgb_depth_fusion.py`)
- Workshop operator → Device benchmark: benchmarks devices (evidence: `README.md`)
- Workshop operator → Local dashboard: starts dashboard (evidence: `README.md`)
- Workshop operator → Responsive dashboard: starts dashboard (evidence: `README.md`)
- OpenVINO model → Local dashboard: loads model (evidence: `README.md`)
- OpenVINO model → Responsive dashboard: loads model (evidence: `README.md`)
- Responsive dashboard → Camera workers: starts workers (evidence: `session_09_integration/local_sdv_pipeline_optimized.py`)
- Responsive dashboard → Detection worker: starts workers (evidence: `session_09_integration/local_sdv_pipeline_optimized.py`)
- Responsive dashboard → Lane worker: starts workers (evidence: `session_09_integration/local_sdv_pipeline_optimized.py`)
- Camera workers → Detection worker: publishes frames (evidence: `session_09_integration/local_sdv_pipeline_optimized.py`)
- Camera workers → Lane worker: provides camera frames (evidence: `session_09_integration/local_sdv_pipeline_optimized.py`)
- Workshop operator → Edge frame streamer: starts remote mode (evidence: `README.md`)
- Edge cameras → Edge frame streamer: provides frames (evidence: `session_09_integration/edge_client_streamer.py`)
- Edge frame streamer → HPC inference server: sends frames (evidence: `session_09_integration/edge_client_streamer.py`)
- HPC inference server → Edge frame streamer: returns annotated frames (evidence: `session_09_integration/hpc_server_inference.py`)
- HPC host → HPC inference server: hosts inference
