mmwavecapture.capture package

Submodules

mmwavecapture.capture.capture module

class mmwavecapture.capture.capture.Capture(base_path: Path)

Bases: object

add_capture_hardware(hw: CaptureHardware) → None

capture() → None

class mmwavecapture.capture.capture.CaptureHardware

Bases: ABC

CaptureHardware is the abstract class for all capture hardware.

For each capture hardware, there are 5 stages of capture process:

1. Initialize capture hardware

   This stage should initialize and config the capture hardware, and make sure the capture hardware is ready to capture data.

   Note

   Do not create any output files in this stage.

2. Prepare capture environment and output files

   This stage should setup the capture environment and create output files for the capture hardware. You should also setup thread/processes for capturing data at this stage, but not start them.

   Warning

   The output files should be created under base_path

   If you are using Capture class, the hardware base_path will be setup after calling Capture.add_capture_hardware(). If you are using CaptureManager class, the hardware base_path will be setup during CaptureManager.capture(). It not using any of the above classes, you should setup the base_path by yourself before calling CaptureHardware.prepare_capture().

3. Start capture

   This stage should start the capture process/thread.

4. Stop capture

   This stage should stop the capture process/thread and close the output files.

5. Dump configuration

   This stage should dump the configuration of the capture hardware to base_path/<config_name> for future reference.

property base_path: Path | None

The base path for the capture hardware

The base path will be set by Capture class or CaptureManager class. Or by yourself if you are not using any of the above classes.

If set it by CaptureManager, the base path should be <dataset_path>/<capture_path>/<hw_name>/.

Setter:

Set the base path for the capture hardware

Getter:

Get the base path for the capture hardware

abstract dump_config() → None

Dump the configuration of the capture hardware to base_path

property hw_name: str

abstract init_capture_hw() → None

Initialize the capture hardware

abstract prepare_capture() → None

Prepare the capture environment and output files

Output filename should be self.base_path/<sensor>.*

abstract start_capture() → None

Start the sensor to capture data

abstract stop_capture() → None

Stop the sensor and capture output files

class mmwavecapture.capture.capture.CaptureManager(config_filename: Path)

Bases: object

Capture Manager manages HDF5-like dataset directory structure and handle capture hardware initialization and capture process.

The layout of dataset directory is as follows:

dataset_path/           # Create when initalizing `CaptureManager`
├── capture_00000/      # Create when calling `CaptureManager.capture()`
│   ├── config.toml     # Capture configuration
│   ├── iwr1843_vert/   # Capture hardware name
│   │   ├── dca.pcap    # DCA1000EVM capture pcap
│   │   ├── radar.cfg   # Radar configuration
│   │   ├── dca.json    # DCA1000EVM configuration
│   ├── realsense/         # Another capture hardware name
│   │   ├── color.avi      # Color video
├── capture_00001/
│   ├── config.toml
│   ├── iwr1843_vert/
│   │   ├── dca.pcap
...

The calling sequence of CaptureManager is as follows:

1. Initialize CaptureManager with config.toml path

2. Initialize capture hardware with CaptureManager.init_hw()

3. Start capture by calling CaptureManager.capture()

CAPTURE_DIR_FORMAT = 'capture_{:05d}'

CAPTURE_DIR_PREFIX = 'capture_'

CAPTURE_LOG_FILENAME = 'capture.log'

CAPTURE_MANAGER_CONFIG_OUTPUT_FILENAME = 'config.toml'

capture() → None

init_hw() → None

mmwavecapture.capture.radardca module

class mmwavecapture.capture.radardca.RadarDCA(hw_name: str, dca_eth_interface: str, radar_config_filename: Path, radar_config_port: str = '/dev/ttyACM0', radar_data_port: str = '/dev/ttyACM1', dca_ip: str = '192.168.33.180', dca_config_port: int = 4096, host_ip: str = '192.168.33.30', capture_frames: int = 100, init_capture_hw: bool = True, **kwargs: Dict[str, Any])

Bases: CaptureHardware

DCA_CONFIG_FILENAME = 'dca.json'

PCAP_OUTPUT_FILENAME = 'dca.pcap'

RADAR_CONFIG_FILENAME = 'radar.cfg'

TCPDUMP_BIN_PATH = '/usr/sbin/tcpdump'

dump_config() → None

Dump the configuration of the capture hardware to base_path

init_capture_hw() → None

Initialize the capture hardware

prepare_capture() → None

Prepare the capture environment and output files

Output filename should be self.base_path/<sensor>.*

start_capture() → None

Start the sensor to capture data

start_tcpdump_capture(outfile: Path) → None

stop_capture() → None

Stop the sensor and capture output files

stop_tcpdump_capture() → None

mmwavecapture.capture.realsense module

class mmwavecapture.capture.realsense.CameraIntrinsics(intrinsics: intrinsics)

Bases: object

Camera intrinsics

Ref: https://intelrealsense.github.io/librealsense/python_docs/_generated/pyrealsense2.intrinsics.html

coeffs: list[float]

Distortion coefficients

fx: float

Focal length of the image plane, as a multiple of pixel width

fy: float

Focal length of the image plane, as a multiple of pixel height

model: str

Distortion model of the image

ppx: float

Horizontal coordinate of the principal point of the image, as a pixel offset from the left edge

ppy: float

Vertical coordinate of the principal point of the image, as a pixel offset from the top edge

class mmwavecapture.capture.realsense.ColorConfig(intrinsics: CameraIntrinsics, fps: int)

Bases: object

fps: int

Frame per second

intrinsics: CameraIntrinsics

Camera intrinsics

class mmwavecapture.capture.realsense.ColorMetadata(frame_num, timestamp, stamp_frame_num, time_of_arrival, backend_timestamp, frame_timestamp, actual_fps)

Bases: NamedTuple

actual_fps: int

Alias for field number 6

backend_timestamp: float

Alias for field number 4

frame_num: int

Alias for field number 0

frame_timestamp: float

Alias for field number 5

stamp_frame_num: int

Alias for field number 2

time_of_arrival: float

Alias for field number 3

timestamp: float

Alias for field number 1

class mmwavecapture.capture.realsense.DepthConfig(intrinsics: CameraIntrinsics, depth_units: float, fps: int, visual_preset: int)

Bases: object

depth_units: float

Depth units

fps: int

Frame per second

intrinsics: CameraIntrinsics

Camera intrinsics

visual_preset: int

Depth visual preset

class mmwavecapture.capture.realsense.DepthMetadata(frame_num, timestamp, stamp_frame_num, time_of_arrival, backend_timestamp, frame_timestamp, actual_fps)

Bases: NamedTuple

actual_fps: int

Alias for field number 6

backend_timestamp: float

Alias for field number 4

frame_num: int

Alias for field number 0

frame_timestamp: float

Alias for field number 5

stamp_frame_num: int

Alias for field number 2

time_of_arrival: float

Alias for field number 3

timestamp: float

Alias for field number 1

class mmwavecapture.capture.realsense.Realsense(hw_name: str, fps: int = 30, resolution: Tuple[int, int] = (1920, 1080), depth_resolution: Tuple[int, int] = (1280, 720), capture_frames: int = 150, rotate: bool = False, latency_skip_frames: int = 3, depth_visual_preset: int = 3, **kwargs: Dict[str, Any])

Bases: CaptureHardware

COLOR_CONFIG_FILENAME = 'color_config.json'

COLOR_METADATA_FILENAME = 'color_metadata.json'

COLOR_OUTPUT_FILENAME = 'color.avi'

DEPTH_CONFIG_FILENAME = 'depth_config.json'

DEPTH_METADATA_FILENAME = 'depth_metadata.json'

DEPTH_OUTPUT_FILENAME = 'depth.zst'

dump_config() → None

Dump the configuration of the capture hardware to base_path

init_capture_hw() → None

Initialize the capture hardware

prepare_capture() → None

Prepare the capture environment and output files

Output filename should be self.base_path/<sensor>.*

start_capture() → None

Start the sensor to capture data

stop_capture() → None

Stop the sensor and capture output files

mmwavecapture.capture.realsense.stamp_framenum(img: ndarray, frame: int) → ndarray

Module contents