utils.plotter.Plotter

utils.plotter.Plotter#

class Plotter(threshold: float = 0.8, show_video: bool = True, save_video: bool = False, file_path: str = 'log', file_name: str = 'out')[source]#

Bases: object

Tool for displaying event videos, predictions and boxes

Parameters:

  • threshold (float, optional) – Threshold value for displaying box. Defaults to 0.8.

  • show_video (bool, optional) – If true, shows video in window. Defaults to True.

  • save_video (bool, optional) – If true, saves the video to a file. Defaults to False.

  • file_path (str, optional) – Folder for saved video. Defaults to “log”.

  • file_name (str, optional) – Save file name. Defaults to “out”.

Methods

apply

Prepares a frame from an event camera for display and overlays prediction and target boxes on it

preprocess

preprocess_events

preprocess_image

__call__(video: List[ndarray], interval: int = 50) → None[source]#

Displays frames obtained by the apply method and saves them

Parameters:

  • video (List[np.ndarray]) – List of frames

  • interval (int) – Time between frames in milliseconds

apply(image: ndarray, predictions: Tensor | None = None, target: Tensor | None = None) → ndarray[source]#

Prepares a frame from an event camera for display and overlays prediction and target boxes on it

Parameters:

  • image (np.ndarray) – Background image

  • predictions (Optional[torch.Tensor]) – Tensor shape [anchor, 6], one label contains (class, iou, xlu, ylu, xrd, yrd).

  • target (Optional[torch.Tensor]) – Ground Truth. Tensor shape [count_box, 5], one label contains (class id, xlu, ylu, xrd, yrd)

Returns:

Returns an image that can be processed by opencv

Return type:

np.ndarray

preprocess(events: Tensor, image: Tensor | None = None)[source]#
preprocess_events(events: Tensor, background: ndarray | None = None) → ndarray[source]#
preprocess_image(img: Tensor) → ndarray[source]#
Contents