from typing import (Sequence, Optional, Union)
import torch
import numpy as np
import matplotlib.colors as mcolors
import matplotlib.patches as mpatches
from rastervision.pytorch_learner.dataset.visualizer import Visualizer # NOQA
from rastervision.pytorch_learner.utils import (
color_to_triple, plot_channel_groups, channel_groups_to_imgs)
[docs]class SemanticSegmentationVisualizer(Visualizer):
"""Plots samples from semantic segmentation Datasets."""
[docs] def plot_xyz(self,
axs: Sequence,
x: torch.Tensor,
y: Union[torch.Tensor, np.ndarray],
z: Optional[torch.Tensor] = None) -> None:
channel_groups = self.get_channel_display_groups(x.shape[1])
img_axes = axs[:len(channel_groups)]
label_ax = axs[len(channel_groups)]
# plot image
imgs = channel_groups_to_imgs(x, channel_groups)
plot_channel_groups(img_axes, imgs, channel_groups)
# plot labels
class_colors = self.class_colors
colors = [
color_to_triple(c) if isinstance(c, str) else c
for c in class_colors
]
colors = np.array(colors) / 255.
cmap = mcolors.ListedColormap(colors)
label_ax.imshow(
y, vmin=0, vmax=len(colors), cmap=cmap, interpolation='none')
label_ax.set_title(f'Ground truth')
label_ax.set_xticks([])
label_ax.set_yticks([])
# plot predictions
if z is not None:
pred_ax = axs[-1]
preds = z.argmax(dim=0)
pred_ax.imshow(
preds,
vmin=0,
vmax=len(colors),
cmap=cmap,
interpolation='none')
pred_ax.set_title(f'Predicted labels')
pred_ax.set_xticks([])
pred_ax.set_yticks([])
# add a legend to the rightmost subplot
class_names = self.class_names
if class_names:
legend_items = [
mpatches.Patch(facecolor=col, edgecolor='black', label=name)
for col, name in zip(colors, class_names)
]
axs[-1].legend(
handles=legend_items,
loc='center right',
bbox_to_anchor=(1.8, 0.5))
[docs] def get_plot_ncols(self, **kwargs) -> int:
x = kwargs['x']
nb_img_channels = x.shape[1]
ncols = len(self.get_channel_display_groups(nb_img_channels)) + 1
z = kwargs.get('z')
if z is not None:
ncols += 1
return ncols