def train_and_validate():
# load data
raw_data = load_digits()
X = raw_data.data
y = raw_data.target
# organize data
X_train = X[:60]
y_train = y[:60]
X_valid = X[60:80]
y_valid = y[60:80]
# train and validate
md_clf = svm_train(X_train, y_train)
y_pred = md_clf.predict(X_valid)
# check the result
idx = [i for i in range(len(y_pred)) if y_valid[i] != y_pred[i]]
for i in idx:
print("index, actual digit, svm_prediction: {:d} {:d} {:d}".format(
i + 60, y_valid[i], y_pred[i]))
precision = (20 - len(idx)) / 20
print("Total number of mid-identifications: {:d}".format(len(idx)))
print("Success rate: {:.2f}".format(precision))
return md_clf
def pred_on_pca():
# load data
raw_data = load_digits()
X = raw_data.data
y = raw_data.target
# organize data
X_train = X[:60]
y_train = y[:60]
X_valid = X[60:80]
y_valid = y[60:80]
# train
pca, X_proj = pca_X(X_train, n_comp=30)
md_clf_new = svm_train(X_proj, y_train)
# validate
file = "unseen_dig.png"
prediction = pca_svm_pred(file, pca, md_clf_new)
print("The prediction: ", prediction[0])
X = prf.load_images(paths, names)
name_dic = {1:" Luke", 2:" Janet", 3:" Gilbert"} #change target dictionary for your own photos and names.
# creating target
y = np.concatenate((np.ones(40), np.ones(37) *2, np.ones(40) * 3)) ## change depending how many targets you will have.
Testing
Errors = 0
for i in range(X.shape[0]):
Xtest = X[i]
ytest = y[i]
Xtrain = np.delete(X, i, axis = 0)
ytrain = np.delete(y, i, axis = 0)
md_pca, Xproj = prf.pca_X(Xtrain, n_comp = 50)
XtestProj = md_pca.transform(Xtest.reshape(1,-1))
md_clf = prf.svm_train(Xproj, ytrain)
predic = md_clf.predict(XtestProj)
if(predic[0] != ytest):
Errors +=1
print("success rate: ", (1 - (Errors/X.shape[0])) * 100, "%")
# START OF PART C
#NOTE MUST CHANGE PICTURE DIRECTORY TO MATCH YOUR INDIVIDUAL DIRECTORY
GuessPicture = "../Pictures/whoswho.JPG"
faces = prf.get_faces(GuessPicture)
predictions = prf.pca_svm_pred(faces, md_pca, md_clf)
print("PCA+SVM predition for person 1: {}".format(name_dic[predictions[0]]))
print("PCA+SVM predition for person 2: {}".format(name_dic[predictions[1]]))
print("PCA+SVM predition for person 3: {}".format(name_dic[predictions[2]]))
im = mpimg.imread('bohr' + str(i) + '.jpeg')
flat_bohr.append(interpol_im(im, dim1=45, dim2=60))
y.append(0)
for i in range(ein):
im = mpimg.imread('ein' + str(i) + '.jpeg')
flat_ein.append(interpol_im(im, dim1=45, dim2=60))
y.append(1)
# concatenate into a 21x2700 array
X = np.vstack((np.array(flat_bohr), np.array(flat_ein)))
###################################### Training ########################################
### instantiating PCA and CLF
md_pca, X_proj = pca_X(X)
md_clf = svm_train(X_proj, y)
### all the training data's predictions are correct
# for i in range(21):
# p = md_clf.predict(X_proj[i].reshape(1, -1))
# print(p, y[i])
###################################### Validation ########################################
print('Below is the Validation part')
perc = 0
tot = 21
for select_idx in range(21):
perc = 0
tot = 21
for select_idx in range(tot):
# set_trace()
Xtrain = np.delete(X, select_idx, axis=0)
ytrain = np.delete(y, select_idx)
md_pca.fit(Xtrain)
Xtrain_proj = md_pca.transform(Xtrain)
Xtest = X[select_idx].reshape(1, -1)
ytest = y[select_idx]
Xtest_proj = md_pca.transform(Xtest)
md_clf = svm_train(Xtrain_proj, ytrain)
pre = md_clf.predict(Xtest_proj)
if pre == ytest:
perc += 1.0
perc /= tot / 100.0
print("Success rate: {:f}".format(perc))
unseen_phys1 = "unseen_phys1.jpg"
unseen_phys2 = "unseen_phys2.jpg"
pre1 = pca_svm_pred(unseen_phys1, md_pca, md_clf)
pre2 = pca_svm_pred(unseen_phys2, md_pca, md_clf)
pre_ein_statement = "PCA+SVM prediction for physicist 1: {:s}".format(phys_dict[pre1[0]])
pre_bohr_statement = "PCA+SVM prediction for physicist 2: {:s}".format(phys_dict[pre2[0]])
dig_img = dig_data.images
unseen = mpimg.imread('unseen_dig.png')
flatten_unseen = interpol_im(unseen, plot_new_im = False)
rescaled_unseen = rescale_pixel(X, flatten_unseen, index, plot_ref = True, plot_unseen = True)
print('Mean pixel value of the rescaled image and the X[{}] image are: {}, {}, repectively'.format(index, np.mean(rescaled_unseen), np.mean(X[15])))
print('The rescaled unseen has all the pixel values as integers, flatten array shown as following:')
print(rescaled_unseen)
###################### Trianing ##########################
dig_data = load_digits()
X = dig_data.data
y = dig_data.target
md_clf = svm_train(X[0:60], y[0:60], gamma = 0.001, C = 100)
####################### Validation #########################
perc = 0
start = 60
end = 80
mis = 0
for i in range(start, end):
pre = md_clf.predict(X[i].reshape(1, -1))[0]
ans = y[i]
# print(pre)
if ans == pre:
perc += 1
else:
plt.imshow(X[i].reshape((8,8)), cmap = 'binary')
print(
'Mean pixel value of the rescaled image and the X[{}] image are: {}, {}, repectively'
.format(index, np.mean(rescaled_unseen), np.mean(X[15])))
print(
'The rescaled unseen has all the pixel values as integers, flatten array shown as following:'
)
print(rescaled_unseen)
###################### Trianing ##########################
dig_data = load_digits()
X = dig_data.data
y = dig_data.target
md_clf = svm_train(X[0:60], y[0:60], gamma=0.001, C=100)
####################### Validation #########################
perc = 0
start = 60
end = 80
mis = 0
for i in range(start, end):
pre = md_clf.predict(X[i].reshape(1, -1))[0]
ans = y[i]
# print(pre)
if ans == pre:
perc += 1
else:
plt.imshow(X[i].reshape((8, 8)), cmap='binary')
from pattern_recog_func import pca_svm_pred
from pdb import set_trace
from scipy.interpolate import interp2d
from skimage import transform, data, io
from sklearn.decomposition import PCA
from sklearn.svm import SVC
from sklearn.datasets import load_digits
from sklearn import preprocessing
from sklearn import svm
dig_data = load_digits()
X = dig_data.data
y = dig_data.target
md_clf = svm_train(X[0:60], y[0:60])
perc = 0
mis = 0
tot_start = 60
tot_end = 80
for i in range(tot_start, tot_end):
ans = y[i]
pre = md_clf.predict(X[i].reshape(1, -1))[0]
if ans != pre:
mis += 1
plt.imshow(X[i].reshape((8, 8)), cmap='binary')
plt.show()
print("--------> index, actual digit, svm_prediction: {:d}, {:d}, {:d}".format(i, ans, pre))
if ans == pre:
perc += 1.
the digits provided
"""
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
from pattern_recog_func import interpol_im
from pattern_recog_func import rescale_pixel
from pattern_recog_func import svm_train
from sklearn.datasets import load_digits
# loading the digits
dig_data = load_digits()
# getting the digits data
X = dig_data.data
y = dig_data.target
# using the first 60 images in X to train
md_clf = svm_train(X[0:60], y[0:60])
perc = 0
mis = 0
tot_start = 60
tot_end = 80
# applying md_clf for the next 20 images
for i in range(tot_start, tot_end):
ans = y[i]
pre = md_clf.predict(X[i].reshape(1, -1))[0]
if ans != pre:
mis += 1
plt.imshow(X[i].reshape((8, 8)), cmap='binary')
plt.show()
print(
"--------> index, actual digit, svm_prediction: {:d}, {:d}, {:d}".
format(i, ans, pre))
im = mpimg.imread('bohr'+str(i)+'.jpeg')
flat_bohr.append(interpol_im(im, dim1 = 45, dim2 = 60))
y.append(0)
for i in range(ein):
im = mpimg.imread('ein'+str(i)+'.jpeg')
flat_ein.append(interpol_im(im, dim1 = 45, dim2 = 60))
y.append(1)
# concatenate into a 21x2700 array
X = np.vstack((np.array(flat_bohr), np.array(flat_ein)))
###################################### Training ########################################
### instantiating PCA and CLF
md_pca, X_proj = pca_X(X)
md_clf = svm_train(X_proj, y)
### all the training data's predictions are correct
# for i in range(21):
# p = md_clf.predict(X_proj[i].reshape(1, -1))
# print(p, y[i])
###################################### Validation ########################################
print('Below is the Validation part')
perc = 0
tot = 21
for select_idx in range(21):