Instructions Write a method classify.py that takes in an image and returns a pre
ID: 3745288 • Letter: I
Question
Instructions
Write a method classify.py that takes in an image and returns a prediction - ball, brick, or cylinder.
An example script in located in challenge/sample_student.py
Your script will be automatically evaluated on a set of test images.
The testing images are quite similar to the training images, and organized into the same difficulty categories.
You are allowed 10 submissions to the evaluation server, which will provide immediate feedback.
The Data
Easy Examples
= this code has syntax error.
Instructions
Write a method classify.py that takes in an image and returns a prediction - ball, brick, or cylinder.
An example script in located in challenge/sample_student.py
Your script will be automatically evaluated on a set of test images.
The testing images are quite similar to the training images, and organized into the same difficulty categories.
You are allowed 10 submissions to the evaluation server, which will provide immediate feedback.
The Data
Easy Examples
= this code has syntax error.
import numpy as np #It's kk to import whatever you want from the local util module if you would like: #from util.X import ... def classify(labels): #method to classify labels random_number = np.randon.randint (low=0, high =len(labels)-1) #finding random integer return labels [random_number] #returning a value print ("Enter some names:") labels = raw_input().split() #splitting labels print ("classify:"), classify(labels) #print generated labelExplanation / Answer
@@@Answer:
"""
These images can be classified using K nearest neighbors
In K nearest neighbor algorithm we just calculate the distance between the
given image and all the training points. And then we will find the nearest training
points which means the nearest labels.
"""
import numpy as np
labels = ['brick', 'ball', 'cylinder']
"""
Brick = 0;
ball = 1;
cylinder = 2;
"""
class KNN(object):
""" a kNN classifier with L2 distance """
def __init__(self):
pass
def train(self, X, y):
"""Training is just easy we just store or memorize the data
"""
self.X_train = X
self.y_train = y
def classify(self, X, k=1):
"""
Predicting the label using trained data
"""
distances = L2Distance(X)
return self.predictLabels(distances)
def L2Distance(self, X):
#Computing distance between Test image and each training point
num_test = X.shape[0]
num_train = self.X_train.shape[0]
dists = np.zeros((num_test, num_train))
dists = np.sqrt(np.sum(X**2, axis=1).reshape(num_test, 1)
+ np.sum(self.X_train**2, axis=1) - 2 * X.dot(self.X_train.T))
return dists
def predictLabels(self, dists, k=1):
"""
We found the distances using L2Distance now we need to find the nearest one to test
image
"""
num_test = dists.shape[0]
y_pred = np.zeros(num_test)
for i in range(num_test):
closest_y = []
top_k_indx = np.argsort(dists[i])[:k]
closest_y = self.y_train[top_k_indx]
vote = Counter(closest_y)
count = vote.most_common()
y_pred[i] = count[0][0]
return y_pred
"""
#See this to have a intuition about how to initialize knn classifer
classifier = KNearestNeighbor()
classifier.train(X_train, y_train)
label = classifier.classify(im)
if im ==0:
out = "brick"
elif(im==1):
out = "ball"
else:
out = "cylinder"
"""