optical.visualizer.visualizer.Visualizer

class optical.visualizer.visualizer.Visualizer(images_dir: Union[str, os.PathLike], dataframe: pandas.core.frame.DataFrame, split: Optional[str] = None, img_size: int = 512, **kwargs)

Bases: object

Creates visualizer to visualze images with annotations by batch size, name and index. Required dataframe of the dataset as input.Can show all images with annotations as a video.

Parameters

• images_dir (Union[str, os.PathLike]) – Path to images in the dataset

• dataframe (pd.DataFrame) – Pandas dataframe which is created by optical.converter. Must contain ["image_id", "x_min", "y_min", "width", "height", "category", "class_id"] columns.

• split (Optional[str], optional) – Split of the dataset to be visualized.

• img_size (int, optional) – Image size to resize and maintain uniformity. Defaults to 512.

__init__(images_dir: Union[str, os.PathLike], dataframe: pandas.core.frame.DataFrame, split: Optional[str] = None, img_size: int = 512, **kwargs)

Methods

__init__(images_dir, dataframe[, split, ...])
reset_filters()	Resets all the filters applied on original dataframe.
show_batch([num_imgs, previous, save_path, ...])	Displays a batch of images based on input size.
show_image([index, name, save_path, render])	Displays images with annotation given index or name.
show_video([use_original])	Displays whole dataset as a video.

reset_filters()

Resets all the filters applied on original dataframe.

show_batch(num_imgs: int = 9, previous: bool = False, save_path: Optional[str] = None, render: str = 'pil', random: bool = True, **kwargs) → Any

Displays a batch of images based on input size.

Parameters

• num_imgs (int, optional) – Number of images and their annotation to be visualized. Defaults to 9.

• previous (bool, optional) – If True just displays last batch. Defaults to False.

• save_path (Optional[str], optional) – Output path if images and annotations to be saved. Defaults to None.

• render (str, optional) – Rendering to be used. Available options are mpl,``pil``,``mpy``. If mpl, uses matplotlib to display the images and annotations. If pil, uses Pillow to display the images and annotations. If mpy, uses mediapy to display as video Defaults to “pil”.

• random (bool, optional) – If True randomly selects num_imgs images otherwise follows a sequence. Defaults to True.

Returns

Incase of Pillow or mediapy rendering IPython media object will be returned.

Return type

Any

show_image(index: int = 0, name: Optional[str] = None, save_path: Optional[str] = None, render: str = 'mpl', **kwargs) → Any

Displays images with annotation given index or name.

Parameters

• index (int, optional) – Index of the image to be fetched. Defaults to 0.

• name (Optional[str], optional) – Name of the image to be fetched. Defaults to None.

• save_path (Optional[str], optional) – Output path if images and annotations to be saved. Defaults to None.

• render (str, optional) – Rendering to be used. Available options are mpl,``pil``,``mpy``. If mpl, uses matplotlib to display the images and annotations. If pil, uses Pillow to display the images and annotations. If mpy, uses mediapy to display as video Defaults to “pil”.

Returns

Incase of Pillow or mediapy rendering IPython media object will be returned.

Return type

Any

show_video(use_original: bool = True, **kwargs) → Any

Displays whole dataset as a video.

Parameters

use_original (bool) – Whether to original dataset or filtered dataset.Defaults to True

Keyword Arguments

• show_image_name (bool) – Whether to show image names in the video or not.

• image_time (float) – How many seconds each should be displayed in the video. e.g: image_time = 1 means each image will be displayed for one second. image_time = 0.5 means each image will be displayed for half a second.

Returns

Returns IPython media object.

Return type

Any