Scribd
Upload
32 views9 pages

CNN - Ipynb - JupyterLab

The document outlines a Python script for training a convolutional neural network (CNN) using TensorFlow and Keras on a dataset of images. It includes data preprocessing steps, model architecture definition, and training process with accuracy and loss metrics over 50 epochs. The dataset consists of 3808 training images and 949 validation images across 9 classes.

Uploaded by

BRUCE WAYNE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
32 views9 pages

CNN - Ipynb - JupyterLab

The document outlines a Python script for training a convolutional neural network (CNN) using TensorFlow and Keras on a dataset of images. It includes data preprocessing steps, model architecture definition, and training process with accuracy and loss metrics over 50 epochs. The dataset consists of 3808 training images and 949 validation images across 9 classes.

Uploaded by

BRUCE WAYNE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

8/3/25, 7:21 PM cnn

In [1]: dataset_path = 'Downloads/dataset/w/RealWaste'

In [3]: import numpy as np


import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dr
from tensorflow.keras.preprocessing.image import ImageDataGenerator

IMG_SIZE = 128
BATCH_SIZE = 32
EPOCHS = 50

In [5]: train_datagen = ImageDataGenerator(


rescale=1./255,
validation_split=0.2,
rotation_range=30,
zoom_range=0.3,
shear_range=0.2,
brightness_range=[0.8, 1.2],
horizontal_flip=True
)

In [6]: train_generator = train_datagen.flow_from_directory(


dataset_path,
target_size=(IMG_SIZE, IMG_SIZE),
batch_size=BATCH_SIZE,
class_mode='categorical',
subset='training'
)

Found 3808 images belonging to 9 classes.

In [10]: val_generator = train_datagen.flow_from_directory(


dataset_path,
target_size=(IMG_SIZE, IMG_SIZE),
batch_size=BATCH_SIZE,
class_mode='categorical',
subset='validation'
)

Found 949 images belonging to 9 classes.

In [12]: model = tf.keras.models.Sequential([


tf.keras.layers.Conv2D(32, (3,3), activation='relu', input_shape=(IMG_SI
tf.keras.layers.MaxPooling2D(2,2),

tf.keras.layers.Conv2D(64, (3,3), activation='relu'),


tf.keras.layers.MaxPooling2D(2,2),

tf.keras.layers.Conv2D(128, (3,3), activation='relu'),


tf.keras.layers.MaxPooling2D(2,2),

tf.keras.layers.Flatten(),

16.16.220.43:8888/lab/tree/cnn.ipynb 1/9
8/3/25, 7:21 PM cnn

tf.keras.layers.Dense(128, activation='relu'),
tf.keras.layers.Dense(9, activation='softmax')
])

In [13]: model.compile(optimizer='adam',
loss='categorical_crossentropy',
metrics=['accuracy'])

from sklearn.utils import class_weight


import numpy as np

# 🧮
Compute class weights
labels = train_generator.classes
class_weights = class_weight.compute_class_weight(
class_weight='balanced',
classes=np.unique(labels),
y=labels
)
class_weights = dict(enumerate(class_weights))

history = model.fit(
train_generator,
validation_data=val_generator,
epochs=EPOCHS,
class_weight=class_weights
)

/home/deepak/anaconda3/lib/python3.11/site-packages/keras/src/trainers/data_
adapters/py_dataset_adapter.py:121: UserWarning: Your `PyDataset` class shou
ld call `super().__init__(**kwargs)` in its constructor. `**kwargs` can incl
ude `workers`, `use_multiprocessing`, `max_queue_size`. Do not pass these ar
guments to `fit()`, as they will be ignored.
self._warn_if_super_not_called()

16.16.220.43:8888/lab/tree/cnn.ipynb 2/9
8/3/25, 7:21 PM cnn

Epoch 1/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 47s 384ms/step - accuracy: 0.1980 - loss: 2.100
2 - val_accuracy: 0.3562 - val_loss: 1.7769
Epoch 2/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 373ms/step - accuracy: 0.4166 - loss: 1.585
7 - val_accuracy: 0.3593 - val_loss: 1.8935
Epoch 3/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 374ms/step - accuracy: 0.4758 - loss: 1.387
4 - val_accuracy: 0.4067 - val_loss: 1.6604
Epoch 4/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 372ms/step - accuracy: 0.5394 - loss: 1.227
5 - val_accuracy: 0.4352 - val_loss: 1.6514
Epoch 5/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 374ms/step - accuracy: 0.5515 - loss: 1.179
9 - val_accuracy: 0.4278 - val_loss: 1.7058
Epoch 6/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 362ms/step - accuracy: 0.5809 - loss: 1.093
1 - val_accuracy: 0.4415 - val_loss: 1.7720
Epoch 7/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 366ms/step - accuracy: 0.5690 - loss: 1.118
6 - val_accuracy: 0.4763 - val_loss: 1.6347
Epoch 8/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 365ms/step - accuracy: 0.6225 - loss: 1.011
8 - val_accuracy: 0.4405 - val_loss: 1.6116
Epoch 9/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 365ms/step - accuracy: 0.6444 - loss: 0.936
6 - val_accuracy: 0.4868 - val_loss: 1.6148
Epoch 10/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 362ms/step - accuracy: 0.6641 - loss: 0.881
5 - val_accuracy: 0.4647 - val_loss: 1.5094
Epoch 11/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 376ms/step - accuracy: 0.6772 - loss: 0.854
8 - val_accuracy: 0.4805 - val_loss: 1.5553
Epoch 12/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 366ms/step - accuracy: 0.7020 - loss: 0.762
0 - val_accuracy: 0.4974 - val_loss: 1.4464
Epoch 13/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 378ms/step - accuracy: 0.6972 - loss: 0.776
3 - val_accuracy: 0.5026 - val_loss: 1.4796
Epoch 14/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 375ms/step - accuracy: 0.7265 - loss: 0.729
4 - val_accuracy: 0.4805 - val_loss: 1.5560
Epoch 15/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 373ms/step - accuracy: 0.7264 - loss: 0.706
2 - val_accuracy: 0.5142 - val_loss: 1.3812
Epoch 16/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 373ms/step - accuracy: 0.7461 - loss: 0.672
4 - val_accuracy: 0.5184 - val_loss: 1.4871
Epoch 17/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 362ms/step - accuracy: 0.7563 - loss: 0.658
9 - val_accuracy: 0.4984 - val_loss: 1.5867
Epoch 18/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 45s 375ms/step - accuracy: 0.7391 - loss: 0.673
2 - val_accuracy: 0.5258 - val_loss: 1.3906
Epoch 19/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 366ms/step - accuracy: 0.7758 - loss: 0.605

16.16.220.43:8888/lab/tree/cnn.ipynb 3/9
8/3/25, 7:21 PM cnn

8 - val_accuracy: 0.5216 - val_loss: 1.3955


Epoch 20/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 356ms/step - accuracy: 0.7697 - loss: 0.611
0 - val_accuracy: 0.5163 - val_loss: 1.5276
Epoch 21/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 359ms/step - accuracy: 0.7863 - loss: 0.560
1 - val_accuracy: 0.5732 - val_loss: 1.4584
Epoch 22/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 358ms/step - accuracy: 0.7865 - loss: 0.572
2 - val_accuracy: 0.5364 - val_loss: 1.5141
Epoch 23/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 356ms/step - accuracy: 0.7891 - loss: 0.544
7 - val_accuracy: 0.5090 - val_loss: 1.4783
Epoch 24/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 41s 347ms/step - accuracy: 0.7952 - loss: 0.516
3 - val_accuracy: 0.5690 - val_loss: 1.4772
Epoch 25/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 368ms/step - accuracy: 0.8126 - loss: 0.498
4 - val_accuracy: 0.5353 - val_loss: 1.6028
Epoch 26/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 368ms/step - accuracy: 0.8199 - loss: 0.462
4 - val_accuracy: 0.4984 - val_loss: 1.7680
Epoch 27/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 356ms/step - accuracy: 0.7916 - loss: 0.542
7 - val_accuracy: 0.5427 - val_loss: 1.6988
Epoch 28/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 365ms/step - accuracy: 0.8148 - loss: 0.497
9 - val_accuracy: 0.5321 - val_loss: 1.5234
Epoch 29/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 362ms/step - accuracy: 0.8163 - loss: 0.499
8 - val_accuracy: 0.5195 - val_loss: 1.6409
Epoch 30/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 366ms/step - accuracy: 0.8337 - loss: 0.430
5 - val_accuracy: 0.5806 - val_loss: 1.5390
Epoch 31/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 361ms/step - accuracy: 0.8176 - loss: 0.471
0 - val_accuracy: 0.5553 - val_loss: 1.6207
Epoch 32/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 364ms/step - accuracy: 0.8453 - loss: 0.361
7 - val_accuracy: 0.5385 - val_loss: 1.7018
Epoch 33/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 358ms/step - accuracy: 0.8463 - loss: 0.406
6 - val_accuracy: 0.5648 - val_loss: 1.6122
Epoch 34/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 357ms/step - accuracy: 0.8423 - loss: 0.367
1 - val_accuracy: 0.5859 - val_loss: 1.5194
Epoch 35/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 355ms/step - accuracy: 0.8517 - loss: 0.375
1 - val_accuracy: 0.5395 - val_loss: 1.6473
Epoch 36/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 354ms/step - accuracy: 0.8516 - loss: 0.373
8 - val_accuracy: 0.5785 - val_loss: 1.6949
Epoch 37/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 361ms/step - accuracy: 0.8576 - loss: 0.356
5 - val_accuracy: 0.6027 - val_loss: 1.4751
Epoch 38/50

16.16.220.43:8888/lab/tree/cnn.ipynb 4/9
8/3/25, 7:21 PM cnn

119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 352ms/step - accuracy: 0.8376 - loss: 0.423


1 - val_accuracy: 0.5901 - val_loss: 1.5955
Epoch 39/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 370ms/step - accuracy: 0.8515 - loss: 0.362
6 - val_accuracy: 0.5353 - val_loss: 1.9740
Epoch 40/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 44s 365ms/step - accuracy: 0.8650 - loss: 0.338
2 - val_accuracy: 0.5111 - val_loss: 1.9862
Epoch 41/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 353ms/step - accuracy: 0.8647 - loss: 0.337
1 - val_accuracy: 0.5848 - val_loss: 1.7867
Epoch 42/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 359ms/step - accuracy: 0.8877 - loss: 0.305
4 - val_accuracy: 0.5869 - val_loss: 1.6642
Epoch 43/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 360ms/step - accuracy: 0.8745 - loss: 0.304
1 - val_accuracy: 0.5638 - val_loss: 1.8951
Epoch 44/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 361ms/step - accuracy: 0.8729 - loss: 0.335
9 - val_accuracy: 0.5669 - val_loss: 1.7508
Epoch 45/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 352ms/step - accuracy: 0.8871 - loss: 0.300
0 - val_accuracy: 0.5574 - val_loss: 1.8962
Epoch 46/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 362ms/step - accuracy: 0.8727 - loss: 0.309
4 - val_accuracy: 0.6101 - val_loss: 1.7305
Epoch 47/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 360ms/step - accuracy: 0.8871 - loss: 0.303
5 - val_accuracy: 0.5638 - val_loss: 1.8268
Epoch 48/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 42s 354ms/step - accuracy: 0.8802 - loss: 0.326
1 - val_accuracy: 0.5933 - val_loss: 1.6941
Epoch 49/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 360ms/step - accuracy: 0.8904 - loss: 0.306
6 - val_accuracy: 0.5911 - val_loss: 1.7019
Epoch 50/50
119/119 ━━━━━━━━━━━━━━━━━━━━ 43s 360ms/step - accuracy: 0.8981 - loss: 0.260
0 - val_accuracy: 0.6217 - val_loss: 1.6943

In [14]: plt.plot(history.history['accuracy'], label='Training Accuracy')


plt.plot(history.history['val_accuracy'], label='Validation Accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.legend()
plt.grid(True)
plt.show()

16.16.220.43:8888/lab/tree/cnn.ipynb 5/9
8/3/25, 7:21 PM cnn

In [15]: import os

📂
# Check your current working directory
print(f" Current working directory: {os.getcwd()}")

# Create a 'models' folder if it doesn't exist


os.makedirs('models', exist_ok=True)

# Save the model

✅
model.save('models/waste_classifier_model.h5')
print(" Model saved successfully in the 'models' folder!")

WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or


`keras.saving.save_model(model)`. This file format is considered legacy. We
recommend using instead the native Keras format, e.g. `model.save('my_model.

📂
keras')` or `keras.saving.save_model(model, 'my_model.keras')`.

✅
Current working directory: /home/deepak
Model saved successfully in the 'models' folder!

✅
In [16]: #model.save('/content/drive/MyDrive/Project/waste_classifier_model.h5')
#print(" Model saved successfully!")

In [17]: import tensorflow as tf

# Load the model from the local 'models' folder

✅
model = tf.keras.models.load_model('models/waste_classifier_model.h5')
print(" Model loaded successfully!")

16.16.220.43:8888/lab/tree/cnn.ipynb 6/9
8/3/25, 7:21 PM cnn

WARNING:absl:Compiled the loaded model, but the compiled metrics have yet to
be built. `model.compile_metrics` will be empty until you train or evaluate

✅
the model.
Model loaded successfully!

In [22]: class_labels = ['Cardboard', 'Food Organics', 'Glass', 'Metal', 'Miscellaneo


'Paper', 'Plastic', 'Textile Trash', 'Vegetation']

In [63]: from tensorflow.keras.preprocessing import image


import numpy as np
import matplotlib.pyplot as plt

# Step 1: Load and preprocess the image

✅
img_path = 'Downloads/dataset/w/RealWaste/Vegetation/Vegetation_193.jpg' #
img = image.load_img(img_path, target_size=(128, 128)) # use same size a
img_array = image.img_to_array(img)
img_array = np.expand_dims(img_array, axis=0)
img_array = img_array / 255.0 # normalize

# Step 2: Predict
prediction = model.predict(img_array)
predicted_index = np.argmax(prediction)
predicted_class = class_labels[predicted_index]

print(" 🔢
Prediction Probabilities:")
for i, label in enumerate(class_labels):
print(f"{label}: {prediction[0][i]*100:.2f}%")

print(f" 🔍 Predicted Waste Type: {predicted_class}")


# Step 3: Show eco-friendly suggestion
eco_tips = {
'Plastic': 'Use reusable cloth or paper bags instead of plastic.',
'Glass': 'Switch to steel or reusable bottles.',
'Food Organics': 'Use compost bins to make fertilizer.',
'Metal': 'Recycle metal items at local scrap centers.',
'Cardboard': 'Reuse or recycle cardboard for packing.',
'Paper': 'Minimize paper use; go digital!',
'Textile Trash': 'Donate usable clothes; avoid fast fashion.',
'Vegetation': 'Compost garden waste.',
'Miscellaneous Trash': 'Avoid disposables; use durable alternatives.'
}

print(f" ♻️ Eco Alternative Tip: {eco_tips[predicted_class]}")

16.16.220.43:8888/lab/tree/cnn.ipynb 7/9
8/3/25, 7:21 PM cnn

🔢
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 39ms/step
Prediction Probabilities:
Cardboard: 0.00%
Food Organics: 0.00%
Glass: 0.00%
Metal: 0.00%
Miscellaneous Trash: 0.00%
Paper: 0.00%
Plastic: 0.00%
Textile Trash: 0.00%

🔍
Vegetation: 100.00%

♻️
Predicted Waste Type: Vegetation
Eco Alternative Tip: Compost garden waste.

In [64]: # Step 4: Map to biodegradability, compostability, recyclability


waste_category_map = {
'Plastic': ['Non-biodegradable', 'Recyclable'],
'Glass': ['Non-biodegradable', 'Recyclable'],
'Food Organics': ['Biodegradable', 'Compostable'],
'Metal': ['Non-biodegradable', 'Recyclable'],
'Cardboard': ['Biodegradable', 'Recyclable'],
'Paper': ['Biodegradable', 'Recyclable'],
'Textile Trash': ['Biodegradable'], # assuming natural fibers
'Vegetation': ['Biodegradable', 'Compostable'],
'Miscellaneous Trash': ['Non-biodegradable']
}

🧾
waste_categories = waste_category_map.get(predicted_class, ['Unknown'])
print(f"\n Waste Category: {', '.join(waste_categories)}")

# Step 5: General waste handling tips


category_tips = {
'Biodegradable': "You can compost this or let it naturally decompose.",
'Non-biodegradable': "Avoid single-use items. Try to reuse or recycle.",
'Compostable': "Put this in a compost bin or compost pile.",
'Recyclable': "Place in a designated recycling bin."
}

print("\n 📌
Waste Handling Tips:")
for cat in waste_categories:
print(f"- {cat}: {category_tips.get(cat, 'No suggestion available.')}")

🧾 Waste Category: Biodegradable, Compostable


📌 Waste Handling Tips:
- Biodegradable: You can compost this or let it naturally decompose.
- Compostable: Put this in a compost bin or compost pile.

In [65]: # Step 4: Display image


plt.imshow(img)
plt.axis('off')
plt.title(f"Predicted: {predicted_class}")
plt.show()

16.16.220.43:8888/lab/tree/cnn.ipynb 8/9
8/3/25, 7:21 PM cnn

In [ ]:

16.16.220.43:8888/lab/tree/cnn.ipynb 9/9

You might also like