I've had a look through and I don't think stack has an answer for this, I am fairly new at this though any help is appreciated.
I'm using an AWS Sagemaker endpoint to return a png mask and I'm trying to display the probability as a whole of each class.
So first stab does this:
np.set_printoptions(threshold=np.inf)
pred_map = np.argmax(mask, axis=0)
non_zero_mask = pred_map[pred_map != 0]) # get everything but background
# print(np.bincount(pred_map[pred_map != 0]).argmax()) # Ignore this line as it just shows the most probable
num_classes = 6
plt.imshow(pred_map, vmin=0, vmax=num_classes-1, cmap='jet')
plt.show()
As you can see I'm removing the background pixels, now I need to show class 1,2,3,4,5 have X probability based on the number of pixels they occupy - I'm unsure if I'll reinvent the wheel by simply taking the total number of elements from the original mask then looping and counting each pixel/class number etc - are there inbuilt methods for this please?
Update:
So after typing this out had a little think and reworded some of searches and came across this.
unique_elements, counts_elements = np.unique(pred_map[pred_map != 0], return_counts=True)
print(np.asarray((unique_elements, counts_elements)))
#[[ 2 3]
#[87430 2131]]
So then I'd just calculate the % based on this or is there a better way? For example I'd do
87430 / 89561(total number of pixels in the mask) * 100
Giving 2 in this case a 97% probability.
Update for Joe's comment below:
rec = Record()
recordio = mx.recordio.MXRecordIO(results_file, 'r')
protobuf = rec.ParseFromString(recordio.read())
values = list(rec.features["target"].float32_tensor.values)
shape = list(rec.features["shape"].int32_tensor.values)
shape = np.squeeze(shape)
mask = np.reshape(np.array(values), shape)
mask = np.squeeze(mask, axis=0)
My first thought was to use np.digitize and write a nice solution.
But then I realized how you can hack it in 10 lines:
import numpy as np
import matplotlib.pyplot as plt
size = (10, 10)
x = np.random.randint(0, 7, size) # your classes, seven excluded.
# empty array, filled with mask and number of occurrences.
x_filled = np.zeros_like(x)
for i in range(1, 7):
mask = x == i
count_mask = np.count_nonzero(mask)
x_filled[mask] = count_mask
print(x_filled)
plt.imshow(x_filled)
plt.colorbar()
plt.show()
I am not sure about the axis convention with imshow
at the moment, you might have to flip the y axis so up is up.
SageMaker does not provide in-built methods for this.
Related
It appears to be impossible to change the y and x axis view limits during an ArtistAnimation, and have the frames replayed with different axis limits.
The limits seem to fixed to those set last before the animation function is called.
In the code below, I have two plotting stages. The input data in the second plot is a much smaller subset of the data in the 1st frame. The data in the 1st stage has a much wider range.
So, I need to "zoom in" when displaying the second plot (otherwise the plot would be very tiny if the axis limits remain the same).
The two plots are overlaid on two different images (that are of the same size, but different content).
import matplotlib
import matplotlib.pyplot as plt
import matplotlib.animation as animation
import matplotlib.image as mpimg
import random
# sample 640x480 image. Actual frame loops through
# many different images, but of same size
image = mpimg.imread('image_demo.png')
fig = plt.figure()
plt.axis('off')
ax = fig.gca()
artists = []
def plot_stage_1():
# both x, y axis limits automatically set to 0 - 100
# when we call ax.imshow with this extent
im_extent = (0, 100, 0, 100) # (xmin, xmax, ymin, ymax)
im = ax.imshow(image, extent=im_extent, animated=True)
# y axis is a list of 100 random numbers between 0 and 100
p, = ax.plot(range(100), random.choices(range(100), k=100))
# Text label at 90, 90
t = ax.text(im_extent[1]*0.9, im_extent[3]*0.9, "Frame 1")
artists.append([im, t, p])
def plot_stage_2():
# axes remain at the the 0 - 100 limit from the previous
# imshow extent so both the background image and plot are tiny
im_extent = (0, 10, 0, 10)
# so let's update the x, y axis limits
ax.set_xlim(im_extent[0], im_extent[1])
ax.set_ylim(im_extent[0], im_extent[3])
im = ax.imshow(image, extent=im_extent, animated=True)
p, = ax.plot(range(10), random.choices(range(10), k=10))
# Text label at 9, 9
t = ax.text(im_extent[1]*0.9, im_extent[3]*0.9, "Frame 2")
artists.append([im, t, p])
plot_stage_1()
plot_stage_2()
# clear white space around plot
fig.subplots_adjust(left=0, bottom=0, right=1, top=1, wspace=None, hspace=None)
# set figure size
fig.set_size_inches(6.67, 5.0, True)
anim = animation.ArtistAnimation(fig, artists, interval=2000, repeat=False, blit=False)
plt.show()
If I call just one of the two functions above, the plot is fine. However, if I call both, the axis limits in both frames will be 0 - 10, 0 - 10. So frame 1 will be super zoomed in.
Also calling ax.set_xlim(0, 100), ax.set_ylim(0, 100) in plot_stage_1() doesn't help. The last set_xlim(), set_ylim() calls fix the axis limits throughout all frames in the animation.
I could keep the axis bounds fixed and apply a scaling function to the input data.
However, I'm curious to know whether I can simply change the axis limits -- my code will be better this way, because the actual code is complicated with multiple stages, zooming plots across many different ranges.
Or perhaps I have to rejig my code to use FuncAnimation, instead of ArtistAnimation?
FuncAnimation appears to result in the expected behavior. So I'm changing my code to use that instead of ArtistAnimation.
Still curious to know though, whether this can at all be done using ArtistAnimation.
I want to normalize the pixel values of an image to the range [0, 1] for each channel (R, G, B).
Minimal Example
#!/usr/bin/env python
import numpy as np
import scipy
from sklearn import preprocessing
original = scipy.misc.imread('Crocodylus-johnsoni-3.jpg')
scipy.misc.imshow(original)
transformed = np.zeros(original.shape, dtype=np.float64)
scaler = preprocessing.MinMaxScaler()
for channel in range(3):
transformed[:, :, channel] = scaler.fit_transform(original[:, :, channel])
scipy.misc.imsave("transformed.jpg", transformed)
What happens
Taking https://commons.wikimedia.org/wiki/File:Crocodylus-johnsoni-3.jpg,
I get the following "normalized" result:
As you can see there are lines from top to bottom at the right side. What happened there? It seems to me that the normalization went wrong. If so: How do I fix it?
In scikit-learn, a two-dimensional array with shape (m, n) is usually interpreted as a collection of m samples, with each sample having n features.
MinMaxScaler.fit_transform() transforms each feature, so each column of your array is transformed independently of the others. That results in the vertical "stripes" in the image.
It looks like you intended to scale each color channel independently. To do that using MinMaxScaler, reshape the input so that each channel becomes one column. That is, if the original image has shape (m, n, 3), reshape it to (m*n, 3) before passing it to the fit_transform() method, and then restore the shape of the result to create the transformed array.
For example,
ascolumns = original.reshape(-1, 3)
t = scaler.fit_transform(ascolumns)
transformed = t.reshape(original.shape)
With this, transformed looks like this:
The image looks exactly like the original, because it turns out that in the array original, the minimum and maximum are 0 and 255, respectively, in each channel:
In [41]: original.min(axis=(0, 1))
Out[41]: array([0, 0, 0], dtype=uint8)
In [42]: original.max(axis=(0, 1))
Out[42]: array([255, 255, 255], dtype=uint8)
So all fit_transform does in this case is transform all the input values to the floating point range [0.0, 1.0] uniformly. If the minimum or maximum was different in one of the channels, the transformed image would look different.
By the way, it is not difficult to perform the transform using pure numpy. (I'm using Python 3, so in the following, the division automatically casts the result to floating point. If you are using Python 2, you'll need to convert one of the argument to floating point, or use from __future__ import division.)
In [58]: omin = original.min(axis=(0, 1), keepdims=True)
In [59]: omax = original.max(axis=(0, 1), keepdims=True)
In [60]: xformed = (original - omin)/(omax - omin)
In [61]: np.allclose(xformed, transformed)
Out[61]: True
(One potential problem with that method is that it will generate an error if one of the channels is constant, because then one of the values in omax - omin will be 0.)
Starting from the Tensorflow CNN example, I'm trying to modify the model to have multiple images as an input (so that the input has not just 3 input channels, but multiples of 3 by stacking images).
To augment the input, I try to use random image operations, such as flipping, contrast and brightness provided in TensorFlow.
My current solution to apply the same random distortion to all input images is to use a fixed seed value for these operations:
def distort_image(image):
flipped_image = tf.image.random_flip_left_right(image, seed=42)
contrast_image = tf.image.random_contrast(flipped_image, lower=0.2, upper=1.8, seed=43)
brightness_image = tf.image.random_brightness(contrast_image, max_delta=0.2, seed=44)
return brightness_image
This method is called multiple times for each image at graph construction time, so I thought for each image it will use the same random number sequence and consequently, it will result in have the same applied image operations for my image input sequence.
# ...
# distort images
distorted_prediction = distort_image(seq_record.prediction)
distorted_input = []
for i in xrange(INPUT_SEQ_LENGTH):
distorted_input.append(distort_image(seq_record.input[i,:,:,:]))
stacked_distorted_input = tf.concat(2, distorted_input)
# Ensure that the random shuffling has good mixing properties.
min_queue_examples = int(num_examples_per_epoch *
MIN_FRACTION_EXAMPLES_IN_QUEUE)
# Generate a batch of sequences and prediction by building up a queue of examples.
return generate_sequence_batch(stacked_distorted_input, distorted_prediction, min_queue_examples,
batch_size, shuffle=True)
In theory, this works fine. And after doing some test runs, this really seemed to solve my problem. But after a while, I found out that I'm having a race-condition, because I use the input pipeline of the CNN-example code with multiple threads (which is the suggested method in TensorFlow to improve performance and reduce memory consumption at runtime):
def generate_sequence_batch(sequence_in, prediction, min_queue_examples,
batch_size):
num_preprocess_threads = 8 # <-- !!!
sequence_batch, prediction_batch = tf.train.shuffle_batch(
[sequence_in, prediction],
batch_size=batch_size,
num_threads=num_preprocess_threads,
capacity=min_queue_examples + 3 * batch_size,
min_after_dequeue=min_queue_examples)
return sequence_batch, prediction_batch
Because multiple threads create my examples, it is not guaranteed anymore that all image operations are performed in the right order (in sense of the right order of random operations).
Here I came to a point where I got completely stuck. Does anyone know how to solve this problem to apply the same image distortion to multiple images?
Some thoughts of mine:
I thought about to do some synchronizations arround these image distortion methods, but I could find anything provided by TensorFlow
I tried to generate to generate a random number for e.g. the random brightness delta using tf.random_uniform() by myself and use this value for tf.image.adjust_contrast(). But the result of the TensorFlow random generator is always a tensor, and I have not found a way to use this tensor as a parameter for tf.image.adjust_contrast() which expects a simple float32 for its contrast_factor parameter.
A solution that would (partly) work would be to combine all images to a huge image using tf.concat(), apply random operations to change contrast and brightness, and split the image afterwards. But this would not work for random flipping, because this would (at least in my case) change the order of the images, and there is no way to detect whether tf.image.random_flip_left_right() has performed a flip or not, which would be required to fix the wrong order of images if necessary.
Here is what I came up with by looking at the code of random_flip_up_down and random_flip_left_right within tensorflow :
def image_distortions(image, distortions):
distort_left_right_random = distortions[0]
mirror = tf.less(tf.pack([1.0, distort_left_right_random, 1.0]), 0.5)
image = tf.reverse(image, mirror)
distort_up_down_random = distortions[1]
mirror = tf.less(tf.pack([distort_up_down_random, 1.0, 1.0]), 0.5)
image = tf.reverse(image, mirror)
return image
distortions = tf.random_uniform([2], 0, 1.0, dtype=tf.float32)
image = image_distortions(image, distortions)
label = image_distortions(label, distortions)
I would do something like this using tf.case. It allows you to specify what to return if certain condition holds https://www.tensorflow.org/api_docs/python/tf/case
import tensorflow as tf
def distort(image, x):
# flip vertically, horizontally, both, or do nothing
image = tf.case({
tf.equal(x,0): lambda: tf.reverse(image,[0]),
tf.equal(x,1): lambda: tf.reverse(image,[1]),
tf.equal(x,2): lambda: tf.reverse(image,[0,1]),
}, default=lambda: image, exclusive=True)
return image
def random_distortion(image):
x = tf.random_uniform([1], 0, 4, dtype=tf.int32)
return distort(image, x[0])
To check if it works.
import numpy as np
import matplotlib.pyplot as plt
# create image
image = np.zeros((25,25))
image[:10,5:10] = 1.
# create subplots
fig, axes = plt.subplots(2,2)
for i in axes.flatten(): i.axis('off')
with tf.Session() as sess:
for i in range(4):
distorted_img = sess.run(distort(image, i))
axes[i % 2][i // 2].imshow(distorted_img, cmap='gray')
plt.show()
I'm plotting a bunch of data on a logscale as a scatter plot in matplotlib (just in case the medium is relevant) using RGBA colors. What I would like to be able to do is once I plotted everything, I want to pick out individual scatter points and change their hue to the hue of some RGB color, but preserving the old alpha value. The way I currently do it is this:
oldHSV = rgb_to_hsv(oldRGBA[:3])
newHSV = rgb_to_hsv(newRGB)
oldHSV[0] = newHSV[0]
newRGBA = hsv_to_rgb(oldHSV).tolist() + [oldRGBA[3]]
i.e. I take the RGB part of the old RGBA value, turn it to HSV, do the same for the new intended RGB color, then replace the hue, turn it back to RGB and add on the old alpha value.
Since I'm doing this thousands of times, this can take significantly longer than I would like to spend here. One possible fix would be to dig into the conversion between RGB and HSV and figure out how to do this in one go, but I was hoping that folks who know how to handle color (I really don't) have figured out simple and efficient ways to do this.
How do I change the hue of a given RGBA color A to that of a given RGB color B while preserving the alpha value of A? Would using a different color model (HSL for example) simplify the task, and if so, which would help?
Here is the solution to do all the replacement in one go :
import matplotlib.colors as clr
import matplotlib.pyplot as plt
import numpy as np
N = 100000
x = 1.2 + 800.0 * np.random.rand(N)
y = 1.2 + 800.0 * np.random.rand(N)
# Generate random colors of the form (r, g, b, a) where r = 0.0
colors = np.random.rand(4 * N).reshape((N, 4))
colors[:, 0] = 0.0
area = np.pi * (5 * np.random.rand(N))**2
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
pcol = ax.scatter(x, y, s=area, c=colors)
ax.set_xscale('log')
ax.set_yscale('log')
# Save initial figure
plt.savefig("hue.jpg")
oldRGBA = pcol.get_facecolors().reshape((N, 1, 4))
oldRGB = oldRGBA[:, :, :3]
newRGB = oldRGB
newRGB[:, :, 0] = 1.0 # Set red component to 1.0
oldHSV = clr.rgb_to_hsv(oldRGB)
newHSV = clr.rgb_to_hsv(newRGB)
oldHSV[:, :, 0] = newHSV[:, :, 0]
newRGBA = np.copy(oldRGBA)
newRGBA[:, :, :3] = clr.rgb_to_hsv(oldHSV)
pcol.set_facecolors(newRGBA[:, 0, :])
# Save modified figure
plt.savefig("hue_bis.jpg")
plt.close()
As you can see, this code attempts to plot 100000 points and in fact it managed to do this in about 2 seconds. Here are the figures produced :
and :
With regard to your last two questions :
How do I change the hue of a given RGBA color A to that of a given RGB color B while preserving the alpha value of A ?
and :
Would using a different color model (HSL for example) simplify the task, and if so, which would help
I think that your approach to do such a modification is appreciable, it avoids making calculations by hand (see HSL and HSV). Using a different color model is possible, both HSL and HSV allow to change the hue without affecting other parameters, but that is only an other way to do it and not a better one.
Hope this will help.
It should be possible to extract the alpha from oldRGBA and apply it to newRGB directly:
newRGBA = colors.to_rgba(newRGB, alpha=oldRGBA[3])
Note: I never used matplotlib, I assume the alpha component is the 4th element of the color
This seems like a fairly straightforward problem, but I'm new to Python and I'm struggling to resolve it. I've got a scatter plot / heatmap generated from two numpy arrays (about 25,000 pieces of information). The y-axis is taken directly from an array and the x-axis is generated from a simple subtraction operation on two arrays.
What I need to do now is slice up the data so that I can work with a selection that falls within certain parameters on the plot. For example, I need to extract all the points that fall within the parallelogram:
I'm able to cut out a rectangle using simple inequalities (see indexing idx_c, idx_h and idx, below) but I really need a way to select the points using a more complex geometry. It looks like this slicing can be done by specifying the vertices of the polygon. This is about the closest I can find to a solution, but I can't figure out how to implement it:
http://matplotlib.org/api/nxutils_api.html#matplotlib.nxutils.points_inside_poly
Ideally, I really need something akin to the indexing below, i.e. something like colorjh[idx]. Ultimately I'll have to plot different quantities (for example, colorjh[idx] vs colorhk[idx]), so the indexing needs to be transferable to all the arrays in the dataset (lots of arrays). Maybe that's obvious, but I would imagine there are solutions that might not be as flexible. In other words, I'll use this plot to select the points I'm interested in, and then I'll need those indices to work for other arrays from the same table.
Here's the code I'm working with:
import numpy as np
from numpy import ndarray
import matplotlib.pyplot as plt
import matplotlib
import atpy
from pylab import *
twomass = atpy.Table()
twomass.read('/IRSA_downloads/2MASS_GCbox1.tbl')
hmag = list([twomass['h_m']])
jmag = list([twomass['j_m']])
kmag = list([twomass['k_m']])
hmag = np.array(hmag)
jmag = np.array(jmag)
kmag = np.array(kmag)
colorjh = np.array(jmag - hmag)
colorhk = np.array(hmag - kmag)
idx_c = (colorjh > -1.01) & (colorjh < 6) #manipulate x-axis slicing here here
idx_h = (hmag > 0) & (hmag < 17.01) #manipulate y-axis slicing here
idx = idx_c & idx_h
# heatmap below
heatmap, xedges, yedges = np.histogram2d(hmag[idx], colorjh[idx], bins=200)
extent = [yedges[0], yedges[-1], xedges[-1], xedges[0]]
plt.clf()
plt.imshow(heatmap, extent=extent, aspect=0.65)
plt.xlabel('Color(J-H)', fontsize=15) #adjust axis labels here
plt.ylabel('Magnitude (H)', fontsize=15)
plt.gca().invert_yaxis() #I put this in to recover familiar axis orientation
plt.legend(loc=2)
plt.title('CMD for Galactic Center (2MASS)', fontsize=20)
plt.grid(True)
colorbar()
plt.show()
Like I say, I'm new to Python, so the less jargon-y the explanation the more likely I'll be able to implement it. Thanks for any help y'all can provide.
a = np.random.randint(0,10,(100,100))
x = np.linspace(-1,5.5,100) # tried to mimic your data boundaries
y = np.linspace(8,16,100)
xx, yy = np.meshgrid(x,y)
m = np.all([yy > xx**2, yy < 10* xx, xx < 4, yy > 9], axis = 0)