add

Doc.py

@@ -0,0 +1,94 @@
+import cv2
+import numpy as np 
+
+## Load image
+img = cv2.imread('Lenna.jpg') ## BGR 
+## Grayscale
+img = cv2.imread('Lenna.jpg',0)
+
+## Show image
+cv2.imshow('image',img)
+cv2.waitKey(0)
+cv2.destroyAllWindows()
+
+## Save image
+cv2.imwrite('test.jpg',img)
+
+## Show image with matplotlib
+import matplotlib.pyplot as plt 
+plt.imshow(img[:,:,::-1]) ## To RGB
+plt.show()
+
+plt.imshow(img,cmap='gray')
+plt.show()
+
+#-----------------Lab1-----------------------
+## Resize
+cv2.resize()
+## Color space 
+cv2.cvtColor()
+## smoothing
+cv2.blur(),cv2.GaussianBlur(),cv2.mediaBlur()
+## PCA Compute
+mean,eigenvectors = cv2.PCACompute(matrix,mean=None)
+
+## min, max
+np.min(), np.max()
+## dot
+np.dot()
+## flatten
+np.flatten()
+
+## KNN
+from sklearn.neighbors import KNeighborsClassifier
+# 宣告
+KNN = KNeighborsClassifier(args)
+# Feed Training Data
+KNN.fit(X_train,Y_train)
+# Predict Class
+KNN.predict(X_test)
+# 直接告訴你acc 
+KNN.score(X_test,Y_test)
+
+#-----------------Lab2-----------------------
+## Opencv
+# ColorMap
+cv2.applyColorMap()
+# Padding
+cv2.copyMakeBorder() #use cv2.BORDER_REFLECT
+# SURF
+cv2.xfeatures2d.SURF_create()
+cv2.drawKeyPoints()
+
+## Numpy scipy
+# Save & load
+np.save(),np.load()
+# concatenate
+np.concatenate()
+# mean 
+np.mean()
+# reciprocal
+np.reciprocal()
+# load .mat file 
+import scipy.io as sio 
+sio.loadmat()
+# Euclidean distance
+from scipy.spatial import distance
+distance.euclidean()
+
+## Sklearn 
+# Kmeans
+from sklearn.cluster import KMeans
+Kmeans = KMeans(args)
+Kmeans.fit_predict(X)
+
+## Matplotlib
+import matplotlib.pyplot as plt
+plt.figure()
+plt.bar(x,y)
+plt.title()
+plt.xlabel()
+plt.ylabel()
+plt.savefig()
+
+

Lab_2/Readme.md

@@ -111,10 +111,12 @@ for simplicity). The centroid of each cluster then indicates a visual word.
 
 	<img src="./image/hardsum.jpg" alt="hs" width="280px"/> <img src="./image/softsum.jpg" alt="ss" width="280px" /> <img src="./image/softmax.jpg" alt="sm" width="280px"/>
 
-4. ) Finally, We adopt the k-nearest neighbors classifier (**KNN**) to perform classification
+  **hint** : You can use `matplotlib.pyplot.bar()` to plot the figure
+
+4. Finally, We adopt the k-nearest neighbors classifier (**KNN**) to perform classification
 using the above BoW features.
 
 	Use Train-100 as the training data and Test-100 for testing (you may choose k = 3
 for simplicity). Report the classification accuracy using **Hard-Sum**, **Soft-Sum**, and
 **Soft-Max**. Are the results as expected (based on your observation on different BoW
-features in 3.)? If not, why?
+features in 3.? If not, why?

Readme.md

@@ -5,6 +5,10 @@ This Python Opencv Lab is split into two parts for morning and afternoon.
 1. [Lab_1/](./Lab_1) : Image Operation, Image Smoothing, Image Denoising, Image PCA & Classification
 2. [Lab_2/](./Lab_2) : Color and Texture Segmentation (Clustering) , Recognition with Bag of Visual Words
 
+### Briefly Introdiction
+
+[Introduction.pdf](./Introduction.pdf)
+
 ### Reference
 The lab practice referred to [OpenCVtutorial](http://opencv-python-tutroals.readthedocs.io/en/latest/index.html) and NTU Deep Learning for Computer Vision (DLCV) course [website](http://vllab.ee.ntu.edu.tw/dlcv.html).