def getScalarMappable(colormap, data=None):
"""Returns matplotlib ScalarMappable corresponding to colormap
:param :class:`.Colormap` colormap: The colormap to convert
:param numpy.ndarray data:
The data on which the colormap is applied.
If provided, it is used to compute autoscale.
:return: matplotlib object corresponding to colormap
:rtype: matplotlib.cm.ScalarMappable
"""
assert colormap is not None
if colormap.getName() is not None:
cmap = getColormap(colormap.getName())
else: # No name, use custom colors
if colormap.getColormapLUT() is None:
raise ValueError(
'addImage: colormap no name nor list of colors.')
colors = colormap.getColormapLUT()
assert len(colors.shape) == 2
assert colors.shape[-1] in (3, 4)
if colors.dtype == numpy.uint8:
# Convert to float in [0., 1.]
colors = colors.astype(numpy.float32) / 255.
cmap = matplotlib.colors.ListedColormap(colors)
vmin, vmax = colormap.getColormapRange(data)
if colormap.getNormalization().startswith('log'):
norm = matplotlib.colors.LogNorm(vmin, vmax)
else: # Linear normalization
norm = matplotlib.colors.Normalize(vmin, vmax)
return matplotlib.cm.ScalarMappable(norm=norm, cmap=cmap)
def getScalarMappable(colormap, data=None):
"""Returns matplotlib ScalarMappable corresponding to colormap
:param :class:`.Colormap` colormap: The colormap to convert
:param numpy.ndarray data:
The data on which the colormap is applied.
If provided, it is used to compute autoscale.
:return: matplotlib object corresponding to colormap
:rtype: matplotlib.cm.ScalarMappable
"""
assert colormap is not None
if colormap.getName() is not None:
cmap = getColormap(colormap.getName())
else: # No name, use custom colors
if colormap.getColormapLUT() is None:
raise ValueError('addImage: colormap no name nor list of colors.')
colors = colormap.getColormapLUT()
assert len(colors.shape) == 2
assert colors.shape[-1] in (3, 4)
if colors.dtype == numpy.uint8:
# Convert to float in [0., 1.]
colors = colors.astype(numpy.float32) / 255.
cmap = matplotlib.colors.ListedColormap(colors)
vmin, vmax = colormap.getColormapRange(data)
if colormap.getNormalization().startswith('log'):
norm = matplotlib.colors.LogNorm(vmin, vmax)
else: # Linear normalization
norm = matplotlib.colors.Normalize(vmin, vmax)
return matplotlib.cm.ScalarMappable(norm=norm, cmap=cmap)
def plot3DHist(hist, means, depth = (3,3,3), colorout = "BGR", filterfunc = lambda x,y,z,h: h > 1000):
bins = (2**depth[0], 2**depth[1], 2**depth[2])
maxsize = (bins[0]*bins[1]*bins[2])
xs = np.zeros(maxsize)
ys = np.zeros(maxsize)
zs = np.zeros(maxsize)
filter = np.zeros(maxsize, dtype = bool)
colors = np.zeros((maxsize,3),dtype= np.float32)
hist = hist.flatten()
# Generate points for indices that exceed a particular threshold
for i in range(maxsize):
z = i % bins[2]
y = (i >> depth[2])%bins[1]
x = (i >> (depth[1]+depth[2])) % bins[0]
colors[i] = np.array([x*256//bins[0],y*256//bins[1],z*256//bins[2]],dtype = np.uint8)
zs[i] = z
ys[i] = y
xs[i] = x
filter[i] = filterfunc(x,y,z,hist[i])
xs = xs[filter]
ys = ys[filter]
zs = zs[filter]
colors = colors[filter]
hist = hist[filter]
if(cvtCode(colorout,"RGB")):
colors = cv.cvtColor(np.array([colors],dtype = np.uint8),cvtCode(colorout,"RGB"))[0]
print(colors)
colors = np.divide(colors,256)
print(colors)
ax1 = plotScatter3D(xs, ys, zs, getColorMap(np.log(1+hist)),bins)
ax2 = plotScatter3D(xs, ys, zs, colors.astype(np.float32),bins)
ax1.scatter(means[:, 0], means[:, 1], means[:, 2], "k", marker = "^")
ax2.scatter(means[:, 0], means[:, 1], means[:, 2], "k", marker = "^")
def getScalarMappable(colormap, data=None):
"""Returns matplotlib ScalarMappable corresponding to colormap
:param :class:`.Colormap` colormap: The colormap to convert
:param numpy.ndarray data:
The data on which the colormap is applied.
If provided, it is used to compute autoscale.
:return: matplotlib object corresponding to colormap
:rtype: matplotlib.cm.ScalarMappable
"""
assert colormap is not None
if colormap.getName() is not None:
cmap = getColormap(colormap.getName())
else: # No name, use custom colors
if colormap.getColormapLUT() is None:
raise ValueError('addImage: colormap no name nor list of colors.')
colors = colormap.getColormapLUT()
assert len(colors.shape) == 2
assert colors.shape[-1] in (3, 4)
if colors.dtype == numpy.uint8:
# Convert to float in [0., 1.]
colors = colors.astype(numpy.float32) / 255.
cmap = matplotlib.colors.ListedColormap(colors)
if colormap.getNormalization().startswith('log'):
vmin, vmax = None, None
if not colormap.isAutoscale():
if colormap.getVMin() > 0.:
vmin = colormap.getVMin()
if colormap.getVMax() > 0.:
vmax = colormap.getVMax()
if vmin is None or vmax is None:
_logger.warning('Log colormap with negative bounds, ' +
'changing bounds to positive ones.')
elif vmin > vmax:
_logger.warning('Colormap bounds are inverted.')
vmin, vmax = vmax, vmin
# Set unset/negative bounds to positive bounds
if vmin is None or vmax is None:
# Convert to numpy array
data = numpy.array(data if data is not None else [], copy=False)
if data.size > 0:
finiteData = data[numpy.isfinite(data)]
posData = finiteData[finiteData > 0]
if vmax is None:
# 1. as an ultimate fallback
vmax = posData.max() if posData.size > 0 else 1.
if vmin is None:
vmin = posData.min() if posData.size > 0 else vmax
if vmin > vmax:
vmin = vmax
else:
vmin, vmax = 1., 1.
norm = matplotlib.colors.LogNorm(vmin, vmax)
else: # Linear normalization
if colormap.isAutoscale():
# Convert to numpy array
data = numpy.array(data if data is not None else [], copy=False)
if data.size == 0:
vmin, vmax = 1., 1.
else:
finiteData = data[numpy.isfinite(data)]
if finiteData.size > 0:
vmin = finiteData.min()
vmax = finiteData.max()
else:
vmin, vmax = 1., 1.
else:
vmin = colormap.getVMin()
vmax = colormap.getVMax()
if vmin > vmax:
_logger.warning('Colormap bounds are inverted.')
vmin, vmax = vmax, vmin
norm = matplotlib.colors.Normalize(vmin, vmax)
return matplotlib.cm.ScalarMappable(norm=norm, cmap=cmap)
def __init__(self, scene_reconstructions, image_files, K, show=True):
self.images = {}
self.num_frames = len(image_files)
frame = 0
for image in image_files:
self.images[frame] = mpimg.imread(image)
frame += 1
self.K = K
self.nH = self.images.itervalues().next().shape[0]
self.nW = self.images.itervalues().next().shape[1]
# prepare point clouds
self.point_clouds = {}
num_objects = len(scene_reconstructions['rotations'])
cmap = get_colors(num_objects + 2)
# cmap = np.array([[1.0000, 0.2857, 0, 1.0],
# [1.0000, 0.5714, 0, 1.0],
# [1.0000, 0.8571, 0, 1.0],
# [0.8571, 1.0000, 0, 1.0],
# [0.5714, 1.0000, 0, 1.0],
# [0, 1.0000, 0.8571, 1.0],
# [0, 0.8571, 1.0000, 1.0],
# [0, 0.5714, 1.0000, 1.0],
# [0, 0.2857, 1.0000, 1.0],
# [0, 0, 1.0000, 1.0],
# [0.2857, 0, 1.0000, 1.0],
# [0.8571, 0, 1.0000, 1.0],
# [0.8571, 0, 0.5000, 1.0],
# [0.5571, 0, 0.5000, 1.0]])
object_index = 0
for label in range(num_objects):
vertices = scene_reconstructions['shapes'][label].T
colors = scene_reconstructions['colors'][label].T
points = scene_reconstructions['points'][label][0]
point_cloud = {}
point_cloud['scale'] = 1.0
# point_cloud['vertices'] = np.ascontiguousarray(vertices)
# point_cloud['colors'] = np.ascontiguousarray(colors)
# point_cloud['vertices'] = vertices.astype(np.float32)
# point_cloud['colors'] = colors.astype(np.float32)
point_cloud['vertices'] = np.ascontiguousarray(
vertices.astype(np.float32))
point_cloud['colors'] = np.ascontiguousarray(
colors.astype(np.float32))
point_cloud['vertex_ids'] = points
point_cloud['seg_color'] = cmap[object_index,
0:3].astype(np.float32)
point_cloud['rots'] = scene_reconstructions['rotations'][label]
point_cloud['trans'] = scene_reconstructions['translations'][label]
self.point_clouds[label] = point_cloud
object_index += 1
# parameters
self.ref_color = cmap[num_objects, 0:3].astype(np.float32)
self.selected_color = cmap[num_objects + 1, 0:3].astype(np.float32)
self.show_ref_color = False
self.show_sel_color = False
self.show_color = True
self.show_gt = False
# self.fix_camera = True
# self.fix_object = -1
self.fix_camera = True
self.fix_object = -1
self.sel_object = -1
self.cur_frame = 0
self.show_camera = False
self.show_all_cameras = False
self.camera_length = 1
self.show_text = False
self.text_size = 12
self.point_size = 5.0
# create canvas
#vispy.scene.SceneCanvas.__init__(self, keys='interactive', show=True)
vispy.scene.SceneCanvas.__init__(self,
keys='interactive',
show=True,
bgcolor='w')
self.unfreeze()
self.view = self.central_widget.add_view()
## add nodes to view
self.create_nodes()
## transform these nodes
self.transform_nodes()
self.view.camera = 'arcball'
# self.view.camera = 'turntable'
if (show):
self.app.run()
