1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: !nvidia-smi

Thu Jan 18 12:06:59 2024

+-------------------------------------------------------------------------
--------------+
| NVIDIA-SMI 535.104.05 Driver Version: 535.104.05 CUDA Vers
ion: 12.2 |
|-----------------------------------------+----------------------+--------
--------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatil
e Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Uti
l Compute M. |
| | |
MIG M. |
|=========================================+======================+========
==============|
| 0 Tesla T4 Off | 00000000:00:04.0 Off |
0 |
| N/A 64C P8 12W / 70W | 0MiB / 15360MiB | 0%
Default |
| | |
N/A |
+-----------------------------------------+----------------------+--------
--------------+

+-------------------------------------------------------------------------
--------------+
| Processes:
|
| GPU GI CI PID Type Process name
GPU Memory |
| ID ID
Usage |
|=========================================================================
==============|
| No running processes found
|
+-------------------------------------------------------------------------
--------------+

In [ ]: !mkdir -p ~/.kaggle
!cp kaggle.json ~/.kaggle/

In [ ]: !kaggle competitions download -c sankalan-crack-detection-challenge

Warning: Your Kaggle API key is readable by other users on this system! To
fix this, you can run 'chmod 600 /root/.kaggle/kaggle.json'
Downloading sankalan-crack-detection-challenge.zip to /content
94% 140M/149M [00:01<00:00, 178MB/s]
100% 149M/149M [00:01<00:00, 144MB/s]

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 1/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: import zipfile
zip_ref=zipfile.ZipFile('/content/sankalan-crack-detection-challenge.zip','
zip_ref.extractall('/content')
zip_ref.close()

In [ ]: import numpy as np
import pandas as pd
import os
import cv2
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow import keras
from PIL import Image
from sklearn.model_selection import train_test_split
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.optimizers import Adam
from sklearn.metrics import accuracy_score
np.random.seed(42)
​
from matplotlib import style
style.use('fivethirtyeight')

In [ ]: train_path='/content/train'
# Resizing the images to 30x30x3
IMG_HEIGHT = 30
IMG_WIDTH = 30
channels = 3

In [ ]: NUM_CATEGORIES = len(os.listdir(train_path))
NUM_CATEGORIES

In [ ]: from google.colab.patches import cv2_imshow

from matplotlib.image import imread

In [ ]: img=plt.imread('/content/train/cracked/1002.jpg')
plt.imshow(img)
plt.grid(visible=None)

In [ ]: os.listdir(train_path)

In [ ]: folders=os.listdir(train_path)
d={}
for i in folders:
class_file=os.listdir(train_path+'/'+i)
d[i]=len(class_file)
d

In [ ]: a=pd.DataFrame(d.items()).rename(columns={0:'Category',1:'Count'})
a

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 2/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: import seaborn as sns

sns.set(rc={'figure.figsize':(10,8)})
sns.barplot(x='Category',y='Count',data=a)
plt.xticks(rotation=90);

In [ ]: imgs=os.listdir('/content/test')
imgs=imgs[:25]
imgs

In [ ]: plt.figure(figsize=(25,25))
for i in range(1,26):
plt.subplot(6,5,i)
img=imread('/content/test/'+imgs[i-1])
plt.imshow(img)
plt.grid(None)

In [ ]: a=cv2.imread('/content/train/cracked/10013.jpg')
a

In [ ]: a=Image.fromarray(a,'RGB')
a

In [ ]: lst=os.listdir('/content/train')
lst

In [ ]: # Here our prime goal is to read the png images in form of array convert th
image_data = []
image_labels = []
for i in lst:
images=os.listdir('/content/train/'+i)
for img in images:
try:
image=cv2.imread('/content/train/'+i+'/'+img)
image_fromarray=Image.fromarray(image,'RGB')
resize_image=image_fromarray.resize((30,30))
image_data.append(np.array(resize_image))
image_labels.append(i)
except:
print('Error in ',img)

In [ ]: zi={'cracked':1,'uncracked':0}
zi

In [ ]: img_labels=[]
for i in image_labels:
if i in zi.keys():
img_labels.append(zi[i])

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 3/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: # Changing the list to numpy array

image_data = np.array(image_data)
img_labels = np.array(img_labels)

In [ ]: #Shuffling the training data

shuffle_indexes = np.arange(image_data.shape[0])
np.random.shuffle(shuffle_indexes)
shuffle_indexes

In [ ]: image_data = image_data[shuffle_indexes]
img_labels = img_labels[shuffle_indexes]

In [ ]: #Splitting the data into train and validation set

X_train, X_val, y_train, y_val = train_test_split(image_data, img_labels, t
​
X_train = X_train/255
X_val = X_val/255

In [ ]: print("X_train.shape", X_train.shape)
print("X_valid.shape", X_val.shape)
print("y_train.shape", y_train.shape)
print("y_valid.shape", y_val.shape)

X_train.shape (10477, 30, 30, 3)

X_valid.shape (4491, 30, 30, 3)
y_train.shape (10477,)
y_valid.shape (4491,)

In [ ]: y_val

Out[27]: array([0, 1, 1, ..., 0, 0, 1])

In [ ]: # One hot encoding the labels

y_train = keras.utils.to_categorical(y_train,2)
y_val = keras.utils.to_categorical(y_val, 2)

In [ ]: print(y_train.shape)
print(y_val.shape)

(10477, 2)
(4491, 2)

In [ ]: #Making the model

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 4/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: model = keras.models.Sequential([
keras.layers.Conv2D(filters=16, kernel_size=(3,3), activation='relu', i
keras.layers.Conv2D(filters=32, kernel_size=(3,3), activation='relu'),
keras.layers.MaxPool2D(pool_size=(2, 2)),
keras.layers.BatchNormalization(axis=-1),
​
keras.layers.Conv2D(filters=64, kernel_size=(3,3), activation='relu'),
keras.layers.Conv2D(filters=128, kernel_size=(3,3), activation='relu'),
keras.layers.MaxPool2D(pool_size=(2, 2)),
keras.layers.BatchNormalization(axis=-1),
​
keras.layers.Flatten(),
keras.layers.Dense(512, activation='relu'),
keras.layers.BatchNormalization(),
keras.layers.Dropout(rate=0.5),
​
keras.layers.Dense(2, activation='sigmoid')
])

In [ ]: learning_rate = 0.001
epochs = 50
lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(
initial_learning_rate=0.001,
decay_steps=10000,
decay_rate=learning_rate/(epochs))
opt = Adam(learning_rate=lr_schedule)
model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['acc

In [ ]: history = model.fit(X_train,y_train, batch_size=32, epochs=epochs, validati

Epoch 1/50
328/328 [==============================] - 11s 10ms/step - loss: 0.9056
- accuracy: 0.5582 - val_loss: 0.6821 - val_accuracy: 0.5736
Epoch 2/50
328/328 [==============================] - 2s 7ms/step - loss: 0.7632 -
accuracy: 0.5700 - val_loss: 0.6755 - val_accuracy: 0.5667
Epoch 3/50
328/328 [==============================] - 3s 9ms/step - loss: 0.7044 -
accuracy: 0.5950 - val_loss: 0.7260 - val_accuracy: 0.5849
Epoch 4/50
328/328 [==============================] - 2s 7ms/step - loss: 0.6742 -
accuracy: 0.6156 - val_loss: 0.7733 - val_accuracy: 0.5674
Epoch 5/50
328/328 [==============================] - 3s 8ms/step - loss: 0.6567 -
accuracy: 0.6352 - val_loss: 0.6802 - val_accuracy: 0.5896
Epoch 6/50
328/328 [==============================] - 2s 7ms/step - loss: 0.6480 -
accuracy: 0.6396 - val_loss: 0.7158 - val_accuracy: 0.5885
Epoch 7/50
328/328 [ ] 3s 8ms/step loss: 0 6303
In [ ]: model.save("model.h5")

/usr/local/lib/python3.10/dist-packages/keras/src/engine/training.py:3103:
UserWarning: You are saving your model as an HDF5 file via `model.save()`.
This file format is considered legacy. We recommend using instead the nati
ve Keras format, e.g. `model.save('my_model.keras')`.
saving_api.save_model(

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 5/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: model.save_weights("/content/output")
​

In [ ]: pd.DataFrame(history.history).plot(figsize=(8, 5))
plt.grid(True)
plt.gca().set_ylim(0, 1)
plt.show()

In [ ]: cat=[i for i in zi.keys()]

vals=[i for i in zi.values()]
zi1=pd.DataFrame(data=[cat,vals]).T.rename(columns={0:'Cat',1:'Vals'})
zi1

Out[37]: Cat Vals

0 cracked 1

1 uncracked 0

In [ ]: data=[]
a=[]
image=cv2.imread('/content/test/1.jpg')
image_fromarray=Image.fromarray(image,'RGB')
resize_image=image_fromarray.resize((30,30))
data.append(np.array(resize_image))
x_test=np.array(data)
x_test=x_test/255
predict=model.predict(x_test)
class_x=np.argmax(predict,axis=1)
print(*class_x)
a.append(*class_x)

1/1 [==============================] - 0s 19ms/step

0

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 6/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: data=[]
pred=[]
for i in os.listdir('/content/test'):
image=cv2.imread('/content/test/'+i)
image_fromarray=Image.fromarray(image,'RGB')
resize_image=image_fromarray.resize((30,30))
data.append(np.array(resize_image))

--------------------------------------------------------------------------
-
NameError Traceback (most recent call las
t)
<ipython-input-59-d05fd46b17b6> in <cell line: 4>()
7 resize_image=image_fromarray.resize((30,30))
8 data.append(np.array(resize_image))
----> 9 label.append(i)

NameError: name 'label' is not defined

In [ ]: x_test=np.array(data)
x_test=x_test/255

In [ ]: predict=model.predict(x_test)
class_x=np.argmax(predict,axis=1)

63/63 [==============================] - 0s 5ms/step

In [ ]: class_x

Out[53]: array([0, 0, 1, ..., 1, 0, 0])

In [ ]: df=pd.DataFrame({'vals':class_x})

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 7/8

1/19/24, 9:39 AM PS4_Ritesh jaiswal_Ritesh_054 - Jupyter Notebook

In [ ]: for i in os.listdir('/content/test'):
df['name']=

Out[55]: vals

0 0

1 0

2 1

3 0

4 1

... ...

1995 1

1996 0

1997 1

1998 0

1999 0

2000 rows × 1 columns

In [ ]: ​

localhost:8888/notebooks/Downloads/Random/COMPETETION/PS4_Ritesh jaiswal_Ritesh_054.ipynb 8/8