In [16]:
import numpy as np
import pandas as pd
import os
import cv2
from PIL import Image
from scipy.misc import imread
import matplotlib.pyplot as plt
import skimage.feature
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelBinarizer
import keras
from keras.models import Sequential, load_model
from keras.layers import Dense, Dropout, Activation, Flatten, Conv2D, MaxPooling2D, Lambda, Cropping2D
from keras.utils import np_utils
from collections import Counter
import datetime
%matplotlib inline
In [17]:
class_names = ['adult_females', 'adult_males', 'juveniles', 'pups', 'subadult_males']
my_dir = "/Volumes/dax/seals/Kaggle-NOAA-SeaLions/"
In [18]:
blacklist_fin = open(my_dir + 'MismatchedTrainImages.txt')
blacklist_ws = blacklist_fin.readlines()
blacklist = []
for i in blacklist_ws:
blacklist.append(i.strip() + '.jpg')
blacklist.append('train.csv')
#print(blacklist[:5])
In [19]:
file_names = os.listdir(my_dir + "Train/")
file_names = sorted(file_names, key=lambda
item: (int(item.partition('.')[0]) if item[0].isdigit() else float('inf'), item))
# select a subset of files to run on
file_names = file_names[0:101]
# dataframe to store results in
coordinates_df = pd.DataFrame(index=file_names, columns=class_names)
In [20]:
for filename in file_names:
file_int = int(filename[:-4])
if file_int%50 == 0:
print('completed %d images' % file_int)
if filename in blacklist:
file_names.remove(filename)
else:
# read the Train and Train Dotted images
image_1 = cv2.imread(my_dir + "/TrainDotted/" + filename)
image_2 = cv2.imread(my_dir + "/Train/" + filename)
cut = np.copy(image_2)
# absolute difference between Train and Train Dotted
image_3 = cv2.absdiff(image_1,image_2)
# mask out blackened regions from Train Dotted
mask_1 = cv2.cvtColor(image_1, cv2.COLOR_BGR2GRAY)
mask_1[mask_1 < 20] = 0
mask_1[mask_1 > 0] = 255
mask_2 = cv2.cvtColor(image_2, cv2.COLOR_BGR2GRAY)
mask_2[mask_2 < 20] = 0
mask_2[mask_2 > 0] = 255
image_3 = cv2.bitwise_or(image_3, image_3, mask=mask_1)
image_3 = cv2.bitwise_or(image_3, image_3, mask=mask_2)
# convert to grayscale to be accepted by skimage.feature.blob_log
image_3 = cv2.cvtColor(image_3, cv2.COLOR_BGR2GRAY)
# detect blobs
blobs = skimage.feature.blob_log(image_3, min_sigma=3, max_sigma=4, num_sigma=1, threshold=0.02)
adult_males = []
subadult_males = []
pups = []
juveniles = []
adult_females = []
image_circles = image_1
for blob in blobs:
# get the coordinates for each blob
y, x, s = blob
# get the color of the pixel from Train Dotted in the center of the blob
g,b,r = image_1[int(y)][int(x)][:]
# decision tree to pick the class of the blob by looking at the color in Train Dotted
if r > 200 and g < 50 and b < 50: # RED
adult_males.append((int(x),int(y)))
cv2.circle(image_circles, (int(x),int(y)), 20, (0,0,255), 10)
elif r > 200 and g > 200 and b < 50: # MAGENTA
subadult_males.append((int(x),int(y)))
cv2.circle(image_circles, (int(x),int(y)), 20, (250,10,250), 10)
elif r < 100 and g < 100 and 150 < b < 200: # GREEN
pups.append((int(x),int(y)))
cv2.circle(image_circles, (int(x),int(y)), 20, (20,180,35), 10)
elif r < 100 and 100 < g and b < 100: # BLUE
juveniles.append((int(x),int(y)))
cv2.circle(image_circles, (int(x),int(y)), 20, (180,60,30), 10)
elif r < 150 and g < 50 and b < 100: # BROWN
adult_females.append((int(x),int(y)))
cv2.circle(image_circles, (int(x),int(y)), 20, (0,42,84), 10)
cv2.rectangle(cut, (int(x)-112,int(y)-112),(int(x)+112,int(y)+112), 0,-1)
coordinates_df["adult_males"][filename] = adult_males
coordinates_df["subadult_males"][filename] = subadult_males
coordinates_df["adult_females"][filename] = adult_females
coordinates_df["juveniles"][filename] = juveniles
coordinates_df["pups"][filename] = pups
%time
In [21]:
f, ax = plt.subplots(1,1,figsize=(10,16))
ax.imshow(cv2.cvtColor(image_circles, cv2.COLOR_BGR2RGB))
plt.show()
In [22]:
f, ax = plt.subplots(1,1,figsize=(10,16))
ax.imshow(cv2.cvtColor(cut, cv2.COLOR_BGR2RGB))
plt.show()
In [23]:
x = []
y = []
In [28]:
coordinates_df[class_names[0]]['4.jpg']
Out[28]:
In [29]:
for filename in file_names:
image = cv2.imread(my_dir + "/Train/" + filename)
for lion_class in class_names:
try:
for coordinates in coordinates_df[lion_class][filename]:
thumb = image[coordinates[1]-32:coordinates[1]+32,coordinates[0]-32:coordinates[0]+32,:]
if np.shape(thumb) == (64, 64, 3):
x.append(thumb)
y.append(lion_class)
except:
pass
In [30]:
for i in range(0,np.shape(cut)[0],224):
for j in range(0,np.shape(cut)[1],224):
thumb = cut[i:i+64,j:j+64,:]
if np.amin(cv2.cvtColor(thumb, cv2.COLOR_BGR2GRAY)) != 0:
if np.shape(thumb) == (64,64,3):
x.append(thumb)
y.append("negative")
In [31]:
class_names.append("negative")
In [32]:
x = np.array(x)
y = np.array(y)
In [33]:
for lion_class in class_names:
f, ax = plt.subplots(1,10,figsize=(12,1.5))
f.suptitle(lion_class)
axes = ax.flatten()
j = 0
for a in axes:
a.set_xticks([])
a.set_yticks([])
for i in range(j,len(x)):
if y[i] == lion_class:
j = i+1
a.imshow(cv2.cvtColor(x[i], cv2.COLOR_BGR2RGB))
break
In [34]:
encoder = LabelBinarizer()
encoder.fit(y)
y = encoder.transform(y).astype(float)
In [35]:
model = Sequential()
model.add(Lambda(lambda x: (x / 255.0) - 0.5, input_shape=(64,64,3)))
model.add(Conv2D(32, (5, 5), activation='relu', padding='same'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Conv2D(64, (5, 5), activation='relu', padding='same'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Conv2D(128, (5, 5), activation='relu', padding='same'))
model.add(Flatten())
model.add(Dense(512, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(6, activation='softmax'))
model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
In [36]:
history = model.fit(x, y, epochs=10, verbose=0)
%time
In [37]:
plt.plot(history.history['acc'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'test'], loc='upper left')
plt.show()
In [38]:
model.save(my_dir + datetime.date.today().isoformat() + '_model.h5')
In [ ]: