COCOVisualizer

class cocohelper.visualizer.COCOVisualizer[source]

Bases: object

This class contains methods to visualize COCO images and annotations.

Parameters:: helper – Coco dataset to visualize.

Method List

`_visualize`(image, annotations[, show_bbox, ...])	Draws an image and its annotations using matplotlib.
`draw_bounding_box`(image, bbox[, color, ...])	Draws a bounding box on top of the input image.
`draw_segmentation_mask`(image, segmentation)	Draw segmentation mask on top of the input image.
`load_image_array`(img_id)	Load image with a given image id and returns it as numpy array.
`pick_color_palette`(number)	Returns standard RGB colors palette (+ random colors, if needed).
`pick_random_colors`(number)	Generate random RGB colors.
`visualize`(img_id[, show_bbox, show_segmentation])	Visualize an image given its image id using matplotlib.

Methods Details

_visualize(image, annotations, show_bbox=False, show_segmentation=False, title=None, img_id=None, figsize=None, dpi=None, ax=None, **kwargs)[source]

Draws an image and its annotations using matplotlib.

Parameters:

image (ndarray) – image to be shown.
annotations (list) – annotations for the given the image.
show_bbox (bool) – if true, show bounding boxes on top of the image.
show_segmentation (bool) – if true, show segmentation masks on top of the image.
title (Optional[str]) – title for the plotted image.
img_id (Optional[int]) – image id in the COCO dataset. Needed only if title is None.
figsize (Optional[Sequence]) – figure size for the plotted image (for details, see matplotlib plt.figure()).
dpi (Optional[int]) – dpi for the plotted image (for details, see matplotlib plt.figure()).
ax (Optional[Axes]) –
optional Axes object, pass one of the axes generated by plt.subplots to show the figure in a subplot. .. rubric:: Examples

To show two figures in two columns >>> fig, axs = plt.subplots(1, 2) >>> ch = COCOHelper(…) >>> visualizer = COCOVisualizer(ch) >>> visualizer.visualize(img_id=0, ax=axs[0]) >>> visualizer.visualize(img_id=1, ax=axs[1]) >>> plt.show()
**kwargs – additional kwargs for matplotlib plt.figure().

Returns:

None.

Return type:

None

static draw_bounding_box(image, bbox, color=(128, 128, 128), title=None, bbox_thickness=2, **kwargs)[source]

Draws a bounding box on top of the input image.

Parameters:

image (ndarray) – numpy array with the image.
bbox (Sequence) – bounding box coordinates.
color (Sequence) – bounding box color, defaults to gray color.
title (Optional[str]) – title for the bounding box, default to empty title.
bbox_thickness (int) – thickness of the bounding box. Use larger number to make the boxes more visible in large images.
**kwargs – optional kwargs for plotting bounding boxes.

Returns:

Numpy array with the given image and a bounding box drawn on top of it.

Return type:

ndarray

static draw_segmentation_mask(image, segmentation, color=(128, 128, 128), **kwargs)[source]

Draw segmentation mask on top of the input image.

Parameters:

image (ndarray) – numpy array with the image.
segmentation (Sequence) – segmentation mask.
color (Sequence) – bounding box color, defaults to gray color.
**kwargs – optional kwargs for plotting the segmentation mask.

Returns:

Numpy array with the given image and a segmentation drawn on top of it.

Return type:

ndarray

load_image_array(img_id)[source]

Load image with a given image id and returns it as numpy array.

Parameters:: img_id (int) – image id.
Returns:: A numpy array with the image associated with the image id in the COCO dataset.
Return type:: ndarray

pick_color_palette(number)[source]

Returns standard RGB colors palette (+ random colors, if needed).

Parameters:: number (int) – the number of RGB color triplets to fetch.
Returns:: A list of triplets with a number RGB colors. The first twelve colors follow a standard palette, the remaining colors will be randomly picked.
Return type:: List

static pick_random_colors(number)[source]

Generate random RGB colors.

Parameters:: number (int) – the number of RGB color triplets to fetch
Returns:: A list of triplets containing a number RGB colors. The colors are randomly picked.
Return type:: List

visualize(img_id, show_bbox=False, show_segmentation=False, **kwargs)[source]

Visualize an image given its image id using matplotlib.

If show_bbox or show_segmentation are True, show also the image annotations on top of the plotted image.

Parameters:

img_id (int) – image id in the COCO dataset.
show_bbox (bool) – if true, show bounding boxes on top of the image.
show_segmentation (bool) – show segmentation masks on top of the image.
**kwargs – additional kwargs for the generated matplotlib figure (see COCOVisualizer._visualize() docs).

Returns:

None.

Return type:

None