Skip to content

imagevisualizer

imagevisualizer

Module for ImageVisualizer

Classes

ImageContainer dataclass

Collection of one or more images, the corresponding prediction and ground truth InstanceLabels and the underlying data of the InstanceLabels.

Functions
get_image_paths 
get_image_paths()

Returns image path(s) as list

Returns:

  • List[str] –

    List of image paths

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
28
29
30
31
32
33
34
35
36
37
def get_image_paths(self) -> List[str]:
    """
    Returns image path(s) as list

    Returns:
        List of image paths
    """
    if isinstance(self.image_path, list):
        return self.image_path
    return [self.image_path]
scale_instance_labels 
scale_instance_labels(scale_factor)

Scales the images prediction and ground truth instance labels with a given scale_factor and returns a new scaled instance of ImageContainer

Parameters:

  • scale_factor (float) –

    Factor to scale the prediction and ground truth labels by

Returns:

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
def scale_instance_labels(self, scale_factor: float) -> "ImageContainer":
    """
    Scales the images prediction and ground truth instance labels with a given
    `scale_factor` and returns a new scaled instance of `ImageContainer`

    Args:
        scale_factor: Factor to scale the prediction and ground truth labels by

    Returns:
        Scaled ImageContainer
    """

    scaled_predictions = [
        prediction.scale_label(scale_factor) for prediction in self.predictions
    ]
    scaled_ground_truth = [
        ground_truth.scale_label(scale_factor) for ground_truth in self.ground_truth
    ]

    return ImageContainer(
        image_path=self.image_path,
        predictions=scaled_predictions,
        ground_truth=scaled_ground_truth,
        visualize_data=self.visualize_data,
        image_visu_size=self.image_visu_size,
        model_output_size=self.model_output_size,
    )

ImageVisualizer 

ImageVisualizer(image_loader, columns_count=2)

Bases: ABC

Visualizer for images of ImageContainers

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
71
72
73
74
75
76
77
def __init__(
    self,
    image_loader: ImageLoader,
    columns_count: int = 2,
):
    self.columns_count = columns_count
    self.image_loader = image_loader
Functions
get_images_with_labels abstractmethod 
get_images_with_labels(image_data_container)

Returns images of 'image_data_container' with drawn prediction and ground truth labels

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
113
114
115
116
117
@abstractmethod
def get_images_with_labels(
    self, image_data_container: ImageContainer
) -> List[np.ndarray]:
    """Returns images of 'image_data_container' with drawn prediction and ground truth labels"""
visualize_images 
visualize_images(image_data_containers)

Visualizes images from 'image_data_containers' with prediction and ground truth labels in streamlit container

Parameters:

  • image_data_containers (List[ImageContainer]) –

    ImageContainer storing information about the images and labels to visualize

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
def visualize_images(  # pylint: disable = too-many-arguments
    self,
    image_data_containers: List[ImageContainer],
):
    """
    Visualizes images from 'image_data_containers' with prediction and ground truth
    labels in streamlit container

    Args:
        image_data_containers: ImageContainer storing information about the images
            and labels to visualize
    """

    for image_data_container in image_data_containers:
        image_data_list = self.get_images_with_labels(image_data_container)
        dashboard_container = st.container()

        columns = dashboard_container.columns(self.columns_count)

        for idx, (image, path) in enumerate(
            zip(image_data_list, image_data_container.get_image_paths())
        ):
            columns[idx % self.columns_count].image(image, caption=basename(path))

        if image_data_container.visualize_data.empty:
            dashboard_container.caption("No predictions found")
        elif len(image_data_list) == 0:
            # pylint:disable = line-too-long
            st.warning(
                f"File not found ({', '.join([basename(path)for path in image_data_container.get_image_paths()])}"
            )
        else:
            dashboard_container.dataframe(image_data_container.visualize_data)

Functions

check_instance_label_type 

check_instance_label_type(label_list, target_type)

Check if each label in label_list is an instance of target_type

Parameters:

  • label_list (List[InstanceLabel]) –

    List of labels for which the type is to be checked

  • target_type (Type[InstanceLabel]) –

    Type to be checked

Returns:

  • bool –

    True if each label of label_list is an instance of target_type,

  • bool –

    otherwise False.

Source code in niceml/dashboard/visualizers/imagevisualizer.py 
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
def check_instance_label_type(
    label_list: List[InstanceLabel], target_type: Type[InstanceLabel]
) -> bool:
    """
    Check if each label in `label_list` is an instance of `target_type`

    Args:
        label_list: List of labels for which the type is to be checked
        target_type: Type to be checked

    Returns:
        True if each label of `label_list` is an instance of `target_type`,
        otherwise False.
    """

    return all(isinstance(label, target_type) for label in label_list)