Struct nokhwa::backends::capture::OpenCvCaptureDevice

pub struct OpenCvCaptureDevice { /* private fields */ }

Available on crate feature input-opencv only.

The backend struct that interfaces with OpenCV. Note that an opencv matching the version that this was either compiled on must be present on the user’s machine. (usually 4.5.2 or greater) For more information, please see opencv-rust and OpenCV VideoCapture Docs.

To see what this does, please see CaptureBackendTrait

Quirks

• Some features don’t work properly on this backend (yet)! Setting Resolution, FPS, FrameFormat does not work and will default to 640x480 30FPS. This is being worked on.
• This is a cross-platform backend. This means that it will work on most platforms given that OpenCV is present.
• This backend can also do IP Camera input.
• The backend’s backend will default to system level APIs on Linux(V4L2), Mac(AVFoundation), and Windows(Media Foundation). Otherwise, it will decide for itself.
• If the OpenCvCaptureDevice is initialized as a IPCamera, the CameraFormat’s index value will be u32::MAX (4294967295).
• OpenCV does not support camera querying. Camera Name and Camera supported resolution/fps/fourcc is a UnsupportedOperationError. Note: resolution(), frame_format(), and frame_rate() is not affected.
• CameraInfo’s human name will be “OpenCV Capture Device {location}”
• CameraInfo’s description will contain the Camera’s Index or IP.
• The API Preference order is the native OS API (linux => v4l2, mac => AVFoundation, windows => MSMF) than CAP_AUTO

Implementations

impl OpenCvCaptureDevice

pub fn new( index: &CameraIndex, cam_fmt: RequestedFormat<'_> ) -> Result<Self, NokhwaError>

Creates a new capture device using the OpenCV backend.

Indexes are gives to devices by the OS, and usually numbered by order of discovery.

IPCameras follow the format

<protocol>://<IP>:<port>/

, but please refer to the manufacturer for the actual IP format.

Errors

If the backend fails to open the camera (e.g. Device does not exist at specified index/ip), Camera does not support specified CameraFormat, and/or other OpenCV Error, this will error.

Panics

If the API u32 -> i32 fails this will error

pub fn is_ip_camera(&self) -> bool

Gets weather said capture device is an IPCamera.

pub fn is_index_camera(&self) -> bool

Gets weather said capture device is an OS-based indexed camera.

pub fn camera_location(&self) -> &CameraIndex

Gets the camera location

pub fn opencv_preference(&self) -> i32

Gets the OpenCV API Preference number. Please refer to OpenCV VideoCapture Flag Docs.

pub fn raw_frame_vec(&mut self) -> Result<Cow<'_, [u8]>, NokhwaError>

Gets the RGB24 frame directly read from OpenCV without any additional processing.

Errors

If the frame is failed to be read, this will error.

pub fn raw_resolution(&self) -> Result<Resolution, NokhwaError>

Gets the resolution raw as read by OpenCV.

Errors

If the resolution is failed to be read (e.g. invalid or not supported), this will error.

pub fn raw_framerate(&self) -> Result<u32, NokhwaError>

Gets the framerate raw as read by OpenCV.

Errors

If the framerate is failed to be read (e.g. invalid or not supported), this will error.

Trait Implementations

impl CaptureBackendTrait for OpenCvCaptureDevice

fn backend(&self) -> ApiBackend

Returns the current backend used.

fn camera_info(&self) -> &CameraInfo

Gets the camera information such as Name and Index as a CameraInfo.

fn refresh_camera_format(&mut self) -> Result<(), NokhwaError>

Forcefully refreshes the stored camera format, bringing it into sync with “reality” (current camera state)

fn camera_format(&self) -> CameraFormat

Gets the current CameraFormat. This will force refresh to the current latest if it has changed.

fn set_camera_format( &mut self, new_fmt: CameraFormat ) -> Result<(), NokhwaError>

Will set the current CameraFormat This will reset the current stream if used while stream is opened.

fn compatible_list_by_resolution( &mut self, _fourcc: FrameFormat ) -> Result<HashMap<Resolution, Vec<u32>>, NokhwaError>

A hashmap of Resolutions mapped to framerates. Not sorted!

fn compatible_fourcc(&mut self) -> Result<Vec<FrameFormat>, NokhwaError>

A Vector of compatible FrameFormats. Will only return 2 elements at most.

fn resolution(&self) -> Resolution

Gets the current camera resolution (See: Resolution, CameraFormat). This will force refresh to the current latest if it has changed.

fn set_resolution(&mut self, new_res: Resolution) -> Result<(), NokhwaError>

Will set the current Resolution This will reset the current stream if used while stream is opened.

fn frame_rate(&self) -> u32

Gets the current camera framerate (See: CameraFormat). This will force refresh to the current latest if it has changed.

fn set_frame_rate(&mut self, new_fps: u32) -> Result<(), NokhwaError>

Will set the current framerate This will reset the current stream if used while stream is opened.

fn frame_format(&self) -> FrameFormat

Gets the current camera’s frame format (See: FrameFormat, CameraFormat). This will force refresh to the current latest if it has changed.

fn set_frame_format(&mut self, fourcc: FrameFormat) -> Result<(), NokhwaError>

Will set the current FrameFormat This will reset the current stream if used while stream is opened.

fn camera_control( &self, control: KnownCameraControl ) -> Result<CameraControl, NokhwaError>

Gets the value of KnownCameraControl.

fn camera_controls(&self) -> Result<Vec<CameraControl>, NokhwaError>

Gets the current supported list of KnownCameraControl

fn set_camera_control( &mut self, id: KnownCameraControl, value: ControlValueSetter ) -> Result<(), NokhwaError>

Sets the control to control in the camera. Usually, the pipeline is calling camera_control(), getting a camera control that way then calling value() to get a ControlValueSetter and setting the value that way.

fn open_stream(&mut self) -> Result<(), NokhwaError>

Will open the camera stream with set parameters. This will be called internally if you try and call frame() before you call open_stream().

fn is_stream_open(&self) -> bool

Checks if stream if open. If it is, it will return true.

fn frame(&mut self) -> Result<Buffer, NokhwaError>

Will get a frame from the camera as a Buffer. Depending on the backend, if you have not called open_stream() before you called this, it will either return an error.

fn frame_raw(&mut self) -> Result<Cow<'_, [u8]>, NokhwaError>

Will get a frame from the camera without any processing applied, meaning you will usually get a frame you need to decode yourself.

fn stop_stream(&mut self) -> Result<(), NokhwaError>

Will drop the stream.

fn compatible_camera_formats( &mut self ) -> Result<Vec<CameraFormat, Global>, NokhwaError>

Gets the compatible CameraFormat of the camera

fn decoded_buffer_size(&self, alpha: bool) -> usize

The minimum buffer size needed to write the current frame. If alpha is true, it will instead return the minimum size of the buffer with an alpha channel as well. This assumes that you are decoding to RGB/RGBA for FrameFormat::MJPEG or FrameFormat::YUYV and Luma8/LumaA8 for FrameFormat::GRAY

fn frame_texture<'a>( &mut self, device: &Device, queue: &Queue, label: Option<&'a str> ) -> Result<Texture, NokhwaError>

Available on crate feature wgpu-types only.

Directly copies a frame to a Wgpu texture. This will automatically convert the frame into a RGBA frame.