Module `src.processing.drawing`

Drawing functions.

These functions use opencv to draw on the provided frames. For performance, they draw in place, mutating the provided image.

Functions

def draw_fps_counter(img: numpy.ndarray, fps: int) ‑> NoneType

Draws fps counter in the bottom left

Args

fps: rounded down fps estimate

def label_bounding_box(img: numpy.ndarray, text: str, orig: Tuple[int, int]) ‑> NoneType

Draws label with text on bounding box in place.

Calculates size of text, draws a background of textsize and draws the input text over it. Both the drawing of background rectangle and text is done in place.

Args

img: image buffer
text: name of gesture
orig: upper left of hand bounding box

def draw_mode_indicator(img: numpy.ndarray, color: str) ‑> NoneType

Draws circular indicator in place.

Based on input bool it sets color as either green or red and draws a circle in the top left corner in place.

Args

img: Image buffer as numpy array.
color: Color of mode indicator, One of {"BLACK", "BLUE", "WHITE", "RED", "GREEN", "GRAY"}.

def bounding_box(img: numpy.ndarray, landmarks: Iterable[mediapipe.framework.formats.landmark_pb2.NormalizedLandmark], gesture: str, padding: int = 15) ‑> NoneType

Draws bounding box in place.

Takes in keypoints and predicted gesture, creates bounding box around hand with extra padding and draws the estimated gesture. Draws in place

Args

img: Image buffer as numpy array.
landmarks: List of hand landmarks for a single hand.
gesture: Name of gesture.
padding: Extra width and height to add outside the outermost points in landmarks

def draw_landmarks(img: numpy.ndarray, hand_landmarks: Iterable[mediapipe.framework.formats.landmark_pb2.NormalizedLandmark]) ‑> NoneType

Draws landmarks on image in place.

Draws joints as circles and the connection of joints as lines on image in place.

Args

img: image buffer
hand_landmarks: List of hand landmarks for a single hand

def draw_slider(img: numpy.ndarray, bounds: Tuple[float, float], slider_pos: float, slider_val: float, value_text: str, orientation: str = 'VERTICAL', size=20) ‑> NoneType

Draws a slider on the image (in place).

Draws a slider with the value text on the image. Can be oriented either horizontally or vertically. The slider will be 'filled' to the slider_val.

Args

img: Image buffer as a np array.
bounds: Normalized coordinates for the bounds, is treated as x or y coordinates depending on HORIZONTAL or VERTICAL orientation respectively.
slider_val: The normalized coordinate for the current slider value. Note: Normalized on the image
value_text: Text to display.
orientation: Orientation for the slider to be drawn, One of "VERTICAL" | "HORIZONTAL"
size: Width/height in pixels depending on orientation being "VERTICAL" or "HORIZONTAL" respectively.

def draw_func_through_points(img: numpy.ndarray, f: Callable[[float], float], normalized_points: List[Tuple[float, float]]) ‑> NoneType

Draws the function f and the provided points in place.

Draws the provided points on the image. Then draws the function f from min x to max x of the provided points. Assumes that f is defined for [x_min, x_max] of the normalized points.

Args

img: Image buffer as a np array.
f: Function to be drawn/interpolate between the points.
normalized_points: List of normalized points [(x0, y0), (x1, y1), …] Can be unsorted.