I am working on Detectron2 and applied the model Instance segmentation on a video by using demo.py
I am trying to add to print the specific class which is sports-ball (class label = 32 on meta data) but I am getting the error after deploying the model. I added this code on the line 177.
for predictions, vis_frame in tqdm.tqdm(demo.run_on_video(video), total=num_frames):
tmp_ = [i for i in predictions['instances'].pred_classes if i.item() == 32]
print('tmp_value is:', tmp_)
if len(tmp_) > 0:
empty_label.append(tmp_[0])
if args.output:
output_file.write(vis_frame)
Related
I am a beginner in machine learning and I am trying to train a model with nltk and tensorflow. But I get the following error when I run my program. I understand the problem. it seems that the shape of my list does not pass but I do not know why and I do not find any relief. I specify that I use a list of lists with different sizes. Need help please I need to understand, solve and move forward
code and error:
I am trying to train a model with nltk and tensorflow. But it seems that the shape of my list does not pass but I do not know why and I do not find any relief. I specify that I use a list of lists with different sizes.
github code:
https://github.com/maeltoukap/whatsapp-chat-bot
First of all, you are missing two reshape steps. You need to add the lines
train_x = np.expand_dims(train_x, axis=1)
train_y = np.expand_dims(train_y, axis=1)
after you define train_x and train_y (so after line 67 in your picture). Your input shape is then the shape of your first training example, so change input_shape: train_x[0] to input_shape: train_x[0].shape. Also change the number of neurons in your last dense layer. Currently you have in your last layer Dense(len(train_y[0]).... You need to change that to Dense(30, ...). Then you should be good.
The complete code would look like this:
random.shuffle(training)
training = np.array(training, dtype=object)
train_x = list(training[:, 0])
train_y = list(training[:, 1])
train_x = np.expand_dims(train_x, axis=1)
train_y = np.expand_dims(train_y, axis=1)
print(len(train_x))
model = Sequential()
# model.add(Dense(128, input_shape=113, activation='relu'))
model.add(Dense(128, input_shape=train_x[0].shape, activation='relu'))
# model.add(Dense(128, input_shape=(len(train_x[0])), activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(54, activation='relu'))
model.add(Dropout(0.5))
# model.add(Dense(113, activation='softmax'))
model.add(Dense(30, activation='softmax'))
sgd = SGD(learning_rate=0.01, momentum=0.9, nesterov=True)
model.compile(optimizer=sgd, loss='categorical_crossentropy', metrics=['accuracy'])
hist = model.fit(train_x, train_y, epochs=200, batch_size=5, verbose=1)
I am reading images from google drive mounted to google colab. I have two folders, one with positive covid-19 chest x-rays and another with normal chest x-rays. I am trying to show these images side-by-side for comparison. Here are images of the code and error:
First Lines Of Code
Error Image
Here is also the written code:
Cimages = ('/content/drive/My Drive/Data/Covid')
Nimages = ('/content/drive/My Drive/Data/Normal')
import skimage
from skimage.transform import resize
def plot(i):
normal = cv2.imread(dataset +'Normal//' + Nimages[i])
normal = skimage.transform.resize(normal, (150,150,3))
covid = cv2.imread(dataset +'Covid//' + Cimages[i])
covid = skimage.transform.resize(normal, (150,150,3), mode = reflect)
pair = np.concatenate((normal, covid), axis = 1)
print('Normal vs. Covid')
plt.figure(figsize=(10,5))
plt.imshow(pair)
plt.show()
for i in range(0,3):
plot(i)
This gives me an error:
AttributeError Traceback (most recent call last)
<ipython-input-52-237aff042641> in <module>()
1 for i in range(0,3):
----> 2 plot(i)
<ipython-input-50-85bb2e03725c> in plot(i)
3 def plot(i):
4 normal = cv2.imread(dataset +'Normal//' + Nimages[i])
----> 5 normal = skimage.transform.resize(normal, (150,150,3))
6 covid = cv2.imread(dataset +'Covid//' + Cimages[i])
7 covid = skimage.transform.resize(normal, (150,150,3), mode = reflect)
/usr/local/lib/python3.6/dist-packages/skimage/transform/_warps.py in resize(image, output_shape, order, mode, cval, clip, preserve_range, anti_aliasing, anti_aliasing_sigma)
89 output_shape = tuple(output_shape)
90 output_ndim = len(output_shape)
---> 91 input_shape = image.shape
92 if output_ndim > image.ndim:
93 # append dimensions to input_shape
AttributeError: 'NoneType' object has no attribute 'shape'
So it seems to be occurring in the skimage.tranform.resize line of code. Please help.
The issue is not with the function
skimage.tranform.resize
but with the reading of the image
normal = cv2.imread(dataset +'Normal//' + Nimages[i])
not sure what are you trying to do there but Nimages[i] won't give you first file in a folder but it will yield first character of a string, in your case /. Then you will send dataset variable + Normal// + / which is basically in your case Normal/// and then you will try to read image on that path, but without doubt the is no image there in which case opencv will return to you None (which is basically nothing). and then you try to resize None with skimage which will fail.
better option would be to read the image directly or in a loop that could look somehow like this:
from os import listdir
from os.path import isfile, join
onlyfiles = [f for f in listdir(Nimages) if isfile(join(Nimages , f))]
for image_path in onlyfiles:
normal = cv2.imread(join(Nimages , image_path))
normal = skimage.transform.resize(normal, (150,150,3))
assuming that there are only images in your mentioned directory.
I'm interested in augmenting my dataset with random image transformations. I'm using Keras ImageDataGenerator, and I'm getting the following error when trying to apply random_transform to a single image:
--> x = apply_transform(x, transform matrix, img_channel_axis, fill_mode, cval)
>>> RuntimeError: affine matrix has wrong number of rows.
I found the source code for the ImageDataGenerator here. However, I'm not sure how to debug the runtime error. Below is the code I have:
from keras.preprocessing.image import img_to_array, load_img
from keras.preprocessing.image import ImageDataGenerator
from keras.applications.inception_v3 import preprocess_input
image_path = './figures/zebra.jpg'
#data augmentation
train_datagen = ImageDataGenerator(
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
print "\nloading image..."
image = load_img(image_path, target_size=(299, 299))
image = img_to_array(image)
image = np.expand_dims(image, axis=0) # 1 x input_shape
image = preprocess_input(image)
train_datagen.fit(image)
image = train_datagen.random_transform(image)
The error occurs at the last line when calling random_transform.
The problem is that random_transform expects a 3D-array.
See the docstring:
def random_transform(self, x, seed=None):
"""Randomly augment a single image tensor.
# Arguments
x: 3D tensor, single image.
seed: random seed.
# Returns
A randomly transformed version of the input (same shape).
"""
So you'll need to call it before np.expand_dims.
I am a beginner in keras and I am trying to classify data with a neural network.
x_train = x_train.reshape(1,x_train.shape[0],window,5)
x_val = x_val.reshape(1,x_val.shape[0],window,5)
x_train = x_train.astype('float32')
x_val = x_val.astype('float32')
model = Sequential()
model.add(Dense(64,activation='relu',input_shape= (data_dim,window,5)))
model.add(Dropout(0.5))
model.add(Dense(64,activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(2,activation='softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='sgd',
metrics=['accuracy'])
weights = model.get_weights()
model_info = model.fit(x_train, y_train,batch_size=batchsize, nb_epoch=15,verbose=1,validation_data=(x_val, y_val))
print x_train.shape
#(1,1600,45,5)
print y_train.shape
#(1600,2)
I always have this error with this script and I don't understand why:
ValueError: Error when checking target: expected dense_3 to have 4 dimensions, but got array with shape (16000, 2)
Your model's output (dense_3, so named because it is the third Dense layer) has four dimensions. However, the labels you are attempting to compare it to (y_train) is only two dimensions. You will need to alter your network's architecture so that your model reshapes the data to match the labels.
Keeping track of tensor shapes is difficult when you're just starting out, so I recommend calling plot_model(model, to_file='model.png', show_shapes=True) before calling model.fit. You can look at the resulting PNG to understand what effect layers are having on the shape of your data.
I am learning python programming and machine learning for my academic project and i found interest in number plate recognition.
By executing below code, i am getting error, which is mentioned below after the code
values=
['0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F','G','H','J','K','L','M','N','P','R','S','T','U','V','W','X','Z']
keys=range(32)
data_map=dict((keys, values))
def get_ann(data_map):
feature_mat=[]
label_mat=[]
for keys in data_map:
path_train="/home/sagar/Project data set/ANPR/ann/%s"%data_map[keys]
filenames=get_imlist(path_train)
perfeature_mat=[]
perlabel_mat=[]
for image in filenames[0]:
raw_image=cv2.imread(image)
raw_image=cv2.cvtColor(raw_image, cv2.COLOR_BGR2GRAY)
#resize the image into 5 cols(width) and 10 rows(height)
raw_image=cv2.resize(raw_image,(5,10), interpolation=cv2.INTER_AREA)
#Do a hard thresholding.
_,th2=cv2.threshold(raw_image, 70, 255, cv2.THRESH_BINARY)
#generate features
horz_hist=np.sum(th2==255, axis=0)
vert_hist=np.sum(th2==255, axis=1)
sample=th2.flatten()
#concatenate these features together
feature=np.concatenate([horz_hist, vert_hist, sample])
# append these features together along with their respective labels
perfeature_mat.append(feature)
perlabel_mat.append(keys)
feature_mat.append(perfeature_mat)
label_mat.append(perlabel_mat)
# These are the final product.
bigfeature_mat=np.vstack(feature_mat)
biglabel_mat=np.hstack(label_mat)
# As usual. We need to convert them into double type for Shogun.
bigfeature_mat=np.array(bigfeature_mat, dtype='double')
biglabel_mat=np.array(biglabel_mat, dtype='double')
#shogun works in a way in which columns are samples and rows are features.
#Hence we need to transpose the observation matrix
obs_matrix=bigfeature_mat.T
#convert the observation matrix and the labels into Shogun RealFeatures and MulticlassLabels structures resp. .
sg_features=RealFeatures(obs_matrix)
sg_labels=MulticlassLabels(biglabel_mat)
#initialize a simple ANN in Shogun with one hidden layer.
layers=DynamicObjectArray()
layers.append_element(NeuralInputLayer(65))
layers.append_element(NeuralLogisticLayer(65))
layers.append_element(NeuralSoftmaxLayer(32))
net=NeuralNetwork(layers)
net.quick_connect()
net.initialize()
net.io.set_loglevel(MSG_INFO)
net.l1_coefficient=3e-4
net.epsilon = 1e-6
net.max_num_epochs = 600
net.set_labels(sg_labels)
net.train(sg_features)
return net
The errors:
AttributeError Traceback (most recent call last)
<ipython-input-28-30225c91fe73> in <module>()
----> 1 net=get_ann(data_map)
<ipython-input-27-809f097ce563> in get_ann(data_map)
59 net=NeuralNetwork(layers)
60 net.quick_connect()
---> 61 net.initialize()
62
63 net.io.set_loglevel(MSG_INFO)
AttributeError: 'NeuralNetwork' object has no attribute 'initialize'
Platform used: Ubuntu 14.04, Python 2.7, opencv-2.4.9, iPython notebook and shogun toolbox.
Can any one please help me in resolving this error? Thanks in advance.
The other code samples are as follows, which have been executed before the above code.
from modshogun import *
def get_vstacked_data(path):
filenames=np.array(get_imlist(path))
#read the image
#convert the image into grayscale.
#change its data-type to double.
#flatten it
vmat=[]
for i in range(filenames[0].shape[0]):
temp=cv2.imread(filenames[0][i])
temp=cv2.cvtColor(temp, cv2.COLOR_BGR2GRAY)
temp=cv2.equalizeHist(temp)
temp=np.array(temp, dtype='double')
temp=temp.flatten()
vmat.append(temp)
vmat=np.vstack(vmat)
return vmat
def get_svm():
#set path for positive training images
path_train='/home/sagar/resized/'
pos_trainmat=get_vstacked_data(path_train)
#set path for negative training images
path_train='/home/sagar/rezize/'
neg_trainmat=get_vstacked_data(path_train)
#form the observation matrix
obs_matrix=np.vstack([pos_trainmat, neg_trainmat])
#shogun works in a way in which columns are samples and rows are features.
#Hence we need to transpose the observation matrix
obs_matrix=obs_matrix.T
#get the labels. Positive training images are marked with +1 and negative with -1
labels=np.ones(obs_matrix.shape[1])
labels[pos_trainmat.shape[0]:obs_matrix.shape[1]]*=-1
#convert the observation matrix and the labels into Shogun RealFeatures and BinaryLabels structures resp. .
sg_features=RealFeatures(obs_matrix)
sg_labels=BinaryLabels(labels)
#Initialise a basic LibSVM in Shogun.
width=2
#kernel=GaussianKernel(sg_features, sg_features, width)
kernel=LinearKernel(sg_features, sg_features)
C=1.0
svm=LibSVM(C, kernel, sg_labels)
_=svm.train()
_=svm.apply(sg_features)
return svm
ocr classification
def validate_ann(cnt):
rect=cv2.minAreaRect(cnt)
box=cv2.cv.BoxPoints(rect)
box=np.int0(box)
output=False
width=rect[1][0]
height=rect[1][1]
if ((width!=0) & (height!=0)):
if (((height/width>1.12) & (height>width)) | ((width/height>1.12) & (width>height))):
if((height*width<1700) & (height*width>100)):
if((max(width, height)<64) & (max(width, height)>35)):
output=True
return output
Probably there are some method deprecations issues with Shogun.
Try to replace:
net.initialize()
With:
net.initialize_neural_network()