Skip to content

README.md

Latest commit

 

History

History
206 lines (146 loc) · 4.27 KB

README.md

File metadata and controls

206 lines (146 loc) · 4.27 KB

Running the CLI

Using uv

The CLI can be run using uv in several ways:

For first, after installing uv,

uv venv
source .venv/bin/activate

uv sync --all-groups
uv run pre-commit install

1. As a Python module

uv run python -m python

With options:

# Use default config.toml
uv run python -m python

# Specify a config file
uv run python -m python --config path/to/config.toml
uv run python -m python -c path/to/config.toml

# Enable debug logging
uv run python -m python --debug

# Run the inference command
uv run python -m python inference
uv run python -m python --debug inference

2. Using the installed script

After installation, you can use the dummy_yolo command:

uv run dummy_yolo

With options:

# Use default config.toml
uv run dummy_yolo

# Specify a config file
uv run dummy_yolo --config path/to/config.toml
uv run dummy_yolo -c path/to/config.toml

# Enable debug logging
uv run dummy_yolo --debug

# Run the inference command
uv run dummy_yolo inference
uv run dummy_yolo --debug inference
uv run dummy_yolo --config config.toml inference

3. Direct module execution

# From the project root
uv run --directory . python -m python

# With options
uv run --directory . python -m python --config config.toml --debug

# Run inference command
uv run --directory . python -m python inference
uv run --directory . python -m python --debug inference

Commands

Default (no command)

Runs the simulation with default behavior.

uv run dummy_yolo

inference

Explicitly runs the inference simulation and sends metadata via ZeroMQ.

uv run dummy_yolo inference
uv run dummy_yolo --debug inference

Options

  • --config, -c: Path to config file (default: config.toml)
  • --debug: Enable debug logging (shows frame-by-frame detection details)

Configuration

The CLI reads settings from config.toml. Example configuration:

[stream]
fps = 25
source_id = 0
uri = "rtsp://camera/stream"

[zmq]
port = 5555

Configuration parameters:

  • stream.fps: Frames per second for simulation (default: 25)
  • stream.source_id: Source identifier for metadata (default: 0)
  • stream.uri: Stream URI for metadata tagging (default: "rtsp://camera/stream")
  • zmq.port: ZeroMQ publisher port (default: 5555)

Metadata Format

The inference simulation sends DeepStream-style metadata via ZeroMQ. Each detection includes:

{
    "uri": "rtsp://camera/stream",
    "class_id": 0,              # Object class (0-2)
    "track_id": 123,            # Unique tracking ID
    "confidence": 0.95,         # Detection confidence
    "bbox": {                   # Bounding box
        "left": 100.0,
        "top": 150.0,
        "width": 50.0,
        "height": 100.0,
        "border_width": 0,
        "has_bg_color": 0
    },
    "frame_num": 42             # Frame number
}

The metadata is published on ZeroMQ with topic b"inference" in the format:

{source_id: [detection1, detection2, ...]}

Example Usage

Running with debug output

uv run dummy_yolo --debug inference

Output:

📡 Sending Live Stream Metadata via ZeroMQ (Ctrl+C to stop)...
📡 Publishing on tcp://*:5555
[Frame 1] Active Objects: 2
  └─ ID 234 (class_id=0) | Confidence: 0.92 | BBox: {'left': 120, 'top': 200, ...}
  └─ ID 456 (class_id=1) | Confidence: 0.88 | BBox: {'left': 300, 'top': 150, ...}
[Frame 2] Active Objects: 3
  └─ ID 234 (class_id=0) | Confidence: 0.91 | BBox: {'left': 122, 'top': 203, ...}
  └─ ID 456 (class_id=1) | Confidence: 0.89 | BBox: {'left': 302, 'top': 148, ...}
  └─ ID 789 (class_id=2) | Confidence: 0.95 | BBox: {'left': 450, 'top': 300, ...}
^C🛑 Stream stopped by user.

Building and Installing

Build the project

uv build

This creates distribution files in the dist/ directory:

  • dist/cpp_refresh-0.1.0-py3-none-any.whl (wheel file)
  • dist/cpp_refresh-0.1.0.tar.gz (source distribution)

Install the wheel file

# Install the wheel file with uv
uv pip install dist/cpp_refresh-0.1.0-py3-none-any.whl

# Or install in editable mode for development
uv pip install -e .

After installation, the dummy_yolo command will be available globally in your environment.

dummy_yolo