def show_image_grid_heatmap(self,
heatmap,
background=None,
ratio=0.3,
normalize=True,
colormap=cm.jet,
name="heatmap",
n_iter=None,
prefix=False,
iter_format="{:05d}",
image_args=None,
**kwargs):
"""
Creates heat map from the given map and if given combines it with the background and then
displays results with as image grid.
Args:
heatmap: 4d- tensor (N, C, H, W) to be converted to a heatmap
background: 4d- tensor (N, C, H, W) background/ context of the heatmap (to be underlayed)
name: The name of the window
ratio: The ratio to mix the map with the background (0 = only background, 1 = only map)
n_iter: The iteration number, formatted with the iter_format and added to the model name (if not None)
iter_format: The format string, which indicates how n_iter will be formated as a string
prefix: If True, the formated n_iter will be appended as a prefix, otherwise as a suffix
image_args: Arguments for the tensorvision save image method
"""
if image_args is None:
image_args = {}
if "normalize" not in image_args:
image_args["normalize"] = normalize
if n_iter is not None:
name = name_and_iter_to_filename(name=name,
n_iter=n_iter,
ending=".png",
iter_format=iter_format,
prefix=prefix)
elif not name.endswith(".png"):
name += ".png"
file_name = os.path.join(self.img_dir, name)
map_grid = np_make_grid(
heatmap, normalize=normalize) # map_grid.shape is (3, X, Y)
if heatmap.shape[1] != 3:
map_ = colormap(map_grid[0])[..., :-1].transpose(2, 0, 1)
else: # heatmap was already RGB, so don't apply colormap
map_ = map_grid
if background is not None:
img_grid = np_make_grid(background, **image_args)
fuse_img = (1.0 - ratio) * img_grid + ratio * map_
else:
fuse_img = map_
fuse_img = np.clip(fuse_img * 255, a_min=0, a_max=255).astype(np.uint8)
imwrite(file_name, fuse_img.transpose(1, 2, 0))
def __show_image_grid_heatmap(self,
heatmap,
tensor=None,
ratio=0.3,
colormap=2,
normalize=True,
name=None,
caption=None,
env_appendix="",
opts=None,
image_args=None,
**kwargs):
"""
Internal show_image_grid_heatmap method, called by the internal process.
This function does all the magic.
"""
from cv2 import cv2
if opts is None: opts = {}
if image_args is None: image_args = {}
map_grid = np_make_grid(heatmap, normalize=normalize)
map_ = np.clip(map_grid * 255, a_min=0, a_max=255)
map_ = map_.astype(np.uint8)
map_ = cv2.applyColorMap(map_.transpose(1, 2, 0), colormap=colormap)
map_ = cv2.cvtColor(map_, cv2.COLOR_BGR2RGB)
map_ = map_.transpose(2, 0, 1)
fuse_img = map_
if tensor is not None:
img_grid = np_make_grid(tensor, **image_args)
image = np.clip(img_grid * 255, a_min=0, a_max=255)
image = image.astype(np.uint8)
fuse_img = (1.0 - ratio) * image + ratio * map_
opts = opts.copy()
opts.update(dict(title=name, caption=caption))
win = self.vis.image(img=fuse_img,
win=name,
env=self.name + env_appendix,
opts=opts)
return win
def show_image_grid(self, image_array, name="Image-Grid", counter=None, nrow=8, padding=2,
normalize=False, range=None, scale_each=False, pad_value=0,
*args, **kwargs):
"""
Sends an array of images to tensorboard as a grid. Like :meth:`.show_image`, but generates
image grid before.
Args:
image_array (np.narray/torch.tensor): Image array/tensor which will be sent as an image grid
name (str): Identifier for the image grid
counter (int): Global step value
nrow (int): Items per row in grid
padding (int): Padding between images in grid
normalize (bool): Normalize images in grid
range (tuple): Tuple (min, max), so images will be normalized to this range
scale_each (bool): If True, each image will be normalized separately instead of using
min and max of whole tensor
pad_value (float): Fill padding with this value
"""
image_args = dict(nrow=nrow,
padding=padding,
normalize=normalize,
range=range,
scale_each=scale_each,
pad_value=pad_value)
if counter is not None:
self.val_dict["{}-image".format(name)] = counter
else:
self.val_dict["{}-image".format(name)] += 1
grid = np_make_grid(image_array, **image_args)
self.writer.add_image(tag=name, img_tensor=grid, global_step=self.val_dict["{}-image".format(name)])
self.val_dict[name] += 1
def show_image_grid_heatmap(self, heatmap, background=None, ratio=0.3, normalize=True,
colormap=cv2.COLORMAP_JET, name="heatmap", n_iter=None,
prefix=False, iter_format="{:05d}", image_args=None, **kwargs):
"""
Creates heat map from the given map and if given combines it with the background and then
displays results with as image grid.
Args:
heatmap: 4d- tensor (N, C, H, W) to be converted to a heatmap
background: 4d- tensor (N, C, H, W) background/ context of the heatmap (to be underlayed)
name: The name of the window
ratio: The ratio to mix the map with the background (0 = only background, 1 = only map)
n_iter: The iteration number, formatted with the iter_format and added to the model name (if not None)
iter_format: The format string, which indicates how n_iter will be formated as a string
prefix: If True, the formated n_iter will be appended as a prefix, otherwise as a suffix
image_args: Arguments for the tensorvision save image method
"""
if image_args is None: image_args = {}
if n_iter is not None:
name = name_and_iter_to_filename(name=name, n_iter=n_iter, ending=".png", iter_format=iter_format,
prefix=prefix)
elif not name.endswith(".png"):
name += ".png"
file_name = os.path.join(self.img_dir, name)
map_grid = np_make_grid(heatmap, normalize=normalize)
map_ = np.clip(map_grid * 255, a_min=0, a_max=255)
map_ = map_.astype(np.uint8)
map_ = cv2.applyColorMap(map_.transpose(1, 2, 0), colormap=colormap)
map_ = cv2.cvtColor(map_, cv2.COLOR_BGR2RGB)
map_ = map_.transpose(2, 0, 1)
fuse_img = map_
if background is not None:
img_grid = np_make_grid(background, **image_args)
image = np.clip(img_grid * 255, a_min=0, a_max=255)
image = image.astype(np.uint8)
fuse_img = (1.0 - ratio) * image + ratio * map_
imsave(file_name, fuse_img.transpose(1, 2, 0))
def show_image_grid(self,
tensor,
name=None,
caption=None,
env_appendix="",
opts=None,
image_args=None,
**kwargs):
"""
Calls the save image grid method (for abstract logger combatibility)
Args:
images: 4d- tensor (N, C, H, W)
name: The name of the window
caption: Caption of the generated image grid
env_appendix: appendix to the environment name, if used the new env is env+env_appendix
opts: opts dict for the ploty/ visdom plot, i.e. can set window size, en/disable ticks,...
image_args: Arguments for the tensorvision save image method
"""
if opts is None: opts = {}
if image_args is None: image_args = {}
if isinstance(tensor, Variable):
tensor = tensor.detach()
if torch.is_tensor(tensor):
assert torch.is_tensor(
tensor), "tensor has to be a pytorch tensor or variable"
assert tensor.dim() == 4, "tensor has to have 4 dimensions"
if not (tensor.size(1) == 1 or tensor.size(1) == 3):
warnings.warn(
"The 1. dimension (channel) has to be either 1 (gray) or 3 (rgb), taking the first "
"dimension now !!!")
tensor = tensor[:, 0:1, ]
grid = make_grid(tensor, **image_args)
image = grid.mul(255).clamp(0, 255).byte().numpy()
elif isinstance(tensor, np.ndarray):
grid = np_make_grid(tensor, **image_args)
image = np.clip(grid * 255, a_min=0, a_max=255)
image = image.astype(np.uint8)
else:
raise ValueError(
"Tensor has to be a torch tensor or a numpy array")
opts = opts.copy()
opts.update(dict(title=name, caption=caption))
win = self.vis.image(img=image,
win=name,
env=self.name + env_appendix,
opts=opts)
return win
def __show_image_grid(self,
tensor,
name=None,
caption=None,
env_appendix="",
opts=None,
image_args=None,
**kwargs):
"""
Internal show_image_grid method, called by the internal process.
This function does all the magic.
"""
if opts is None: opts = {}
if image_args is None: image_args = {}
if isinstance(tensor, Variable):
tensor = tensor.detach()
if torch.is_tensor(tensor):
assert torch.is_tensor(
tensor), "tensor has to be a pytorch tensor or variable"
assert tensor.dim() == 4, "tensor has to have 4 dimensions"
if not (tensor.size(1) == 1 or tensor.size(1) == 3):
warnings.warn(
"The 1. dimension (channel) has to be either 1 (gray) or 3 (rgb), taking the first "
"dimension now !!!")
tensor = tensor[:, 0:1, ]
grid = make_grid(tensor, **image_args)
image = grid.mul(255).clamp(0, 255).byte().numpy()
elif isinstance(tensor, np.ndarray):
grid = np_make_grid(tensor, **image_args)
image = np.clip(grid * 255, a_min=0, a_max=255)
image = image.astype(np.uint8)
else:
raise ValueError(
"Tensor has to be a torch tensor or a numpy array")
opts = opts.copy()
opts.update(dict(title=name, caption=caption))
win = self.vis.image(img=image,
win=name,
env=self.name + env_appendix,
opts=opts)
return win
def show_image_grid_heatmap(self,
heatmap,
background=None,
ratio=0.3,
colormap=cm.jet,
normalize=True,
name="heatmap",
caption=None,
env_appendix="",
opts=None,
image_args=None,
**kwargs):
"""
Creates heat map from the given map and if given combines it with the background and then
displays results with as image grid.
Args:
heatmap: 4d- tensor (N, C, H, W), if C = 3, colormap won't be applied.
background: 4d- tensor (N, C, H, W)
name: The name of the window
ratio: The ratio to mix the map with the background (0 = only background, 1 = only map)
caption: Caption of the generated image grid
env_appendix: appendix to the environment name, if used the new env is env+env_appendix
opts: opts dict for the ploty/ visdom plot, i.e. can set window size, en/disable ticks,...
image_args: Arguments for the tensorvision save image method
"""
if opts is None:
opts = {}
if image_args is None:
image_args = {}
if "normalize" not in image_args:
image_args["normalize"] = normalize
# if len(heatmap.shape) != 4:
# raise IndexError("'heatmap' must have dimensions BxCxHxW!")
map_grid = np_make_grid(
heatmap, normalize=normalize) # map_grid.shape is (3, X, Y)
if heatmap.shape[1] != 3:
map_ = colormap(map_grid[0])[..., :-1].transpose(2, 0, 1)
else: # heatmap was already RGB, so don't apply colormap
map_ = map_grid
if background is not None:
img_grid = np_make_grid(background, **image_args)
fuse_img = (1.0 - ratio) * img_grid + ratio * map_
else:
fuse_img = map_
fuse_img = np.clip(fuse_img * 255, a_min=0, a_max=255).astype(np.uint8)
opts = opts.copy()
opts.update(dict(title=name, caption=caption))
win = self.vis.image(img=fuse_img,
win=name,
env=self.name + env_appendix,
opts=opts)
return win