cv2_visualizer

class real_robot.utils.visualization.cv2_visualizer.CV2Visualizer(window_name='Images', *, update_drawing_fn: Callable[[ndarray, dict[str, ndarray]], tuple[ndarray, Any]] | None = None, run_as_process=False, stream_camera=False)

Bases: object

OpenCV visualizer for RGB and depth images, fps is updated in window title

class DrawingMode(value)

Bases: Enum

Drawing mode when draw_with_mouse()

Box = 0

Point = 1

box_labels: np.ndarray = array([], dtype=int64)

boxes: np.ndarray = array([], shape=(0, 4), dtype=int64)

call_update_drawing_fn() → None

Call self._update_drawing_fn with current drawings This happens when the user adds/removes any drawing.

clear_image()

Show a black image

close()

draw_with_mouse(*, update_drawing_fn: Callable[[ndarray, dict[str, ndarray]], tuple[ndarray, Any]] | None = None) → tuple[ndarray, dict[str, ndarray], Any] | None

Enter drawing mode with mouse

Parameters:

update_drawing_fn – function to be called with the image and drawn points/boxes. Inputs to this function are an RGB image and a dictionary of {“points”: np.ndarray, “point_labels”: np.ndarray, “boxes”: np.ndarray, “box_labels”: np.ndarray} Outputs are the modified image and any additional data.

Returns:

Original image before drawing, drawn labels dict, _extra_ret_data returned from self._update_drawing_fn()

property drawn_labels: dict[str, ndarray]

static get_image_layout(n_image: int) → tuple[int, int]

Get layout of Images (n_rows, n_cols) where n_rows >= n_cols

image_boundaries: np.ndarray = array([], shape=(0, 2, 2), dtype=float64)

images: list[np.ndarray] = []

on_mouse(event: int, x: int, y: int, flags: int, param: Any | None = None)

Callback function for mouse events.

Parameters:

• event – one of the cv2.MouseEventTypes: [EVENT_MOUSEMOVE, EVENT_LBUTTONDOWN, EVENT_RBUTTONDOWN, EVENT_MBUTTONDOWN, EVENT_LBUTTONUP, EVENT_RBUTTONUP, EVENT_MBUTTONUP, EVENT_LBUTTONDBLCLK, EVENT_RBUTTONDBLCLK, EVENT_MBUTTONDBLCLK, EVENT_MOUSEWHEEL, EVENT_MOUSEHWHEEL] https://docs.opencv.org/4.9.0/d0/d90/group__highgui__window__flags.html#ga927593befdddc7e7013602bca9b079b0

• x – The x-coordinate of the mouse event.

• y – The y-coordinate of the mouse event.

• flags – one of the cv2.MouseEventFlags: [EVENT_FLAG_LBUTTON, EVENT_FLAG_RBUTTON, EVENT_FLAG_MBUTTON, EVENT_FLAG_CTRLKEY, EVENT_FLAG_SHIFTKEY, EVENT_FLAG_ALTKEY] https://docs.opencv.org/4.9.0/d0/d90/group__highgui__window__flags.html#gaab4dc057947f70058c80626c9f1c25ce

• param – The optional user data passed by cv2.setMouseCallback().

point_labels: np.ndarray = array([], dtype=int64)

points: np.ndarray = array([], shape=(0, 2), dtype=int64)

static preprocess_image(image: ndarray, depth_scale=1000.0) → ndarray

Preprocess image for plotting with cv2 (RGB2BGR, applyColorMap for depth) :param image: depth, grayscale or RGB color image :param depth_scale: used to apply color map on np.floating depth_image :return image: color image in BGR format, [H, W, 3] np.uint8 np.ndarray

render()

Update renderer to show image and respond to mouse and keyboard events

static resize_image(image: ndarray, max_H: int, max_W: int, interpolation=6) → ndarray

Resize image to the requested size while preserving its aspect ratio

run_as_process()

Run CV2Visualizer as a separate process

selected_image_idx: int | None = None

show_images(images: list[ndarray])

Show the list of images, support non-equal size (cv2.rerize to max size)

Parameters:

images – List of np.ndarray images. Supports depth or RGB color images. If depth image, dtype can be np.uint16 or np.floating If RGB image, dtype must be np.uint8