Scribd
Upload
11 views

Program 2 (C)

The document describes designing and analyzing a CNN architecture for object recognition. It discusses transferring learning from a pre-trained VGG16 model on ImageNet to a new model for classifying objects in CIFAR-10. The algorithm loads the VGG16 model without its top layers, adds new top layers, compiles and trains the model on CIFAR-10 for 10 epochs. It evaluates the model's ability to perform object recognition via transfer learning from the pre-trained weights.

Uploaded by

Rahul Rahul
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
11 views

Program 2 (C)

The document describes designing and analyzing a CNN architecture for object recognition. It discusses transferring learning from a pre-trained VGG16 model on ImageNet to a new model for classifying objects in CIFAR-10. The algorithm loads the VGG16 model without its top layers, adds new top layers, compiles and trains the model on CIFAR-10 for 10 epochs. It evaluates the model's ability to perform object recognition via transfer learning from the pre-trained weights.

Uploaded by

Rahul Rahul
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Experiment No: 02. (C) Date: ……. /…….

/……………

DESIGN AND ANALYSIS OF A CNN ARCHITECTURE FOR OBJECT


RECOGNITION

Aim

To write a python program to analyse the CNN architecture for object recognition.

Equipment Required

PC with python software

Algorithm

Step 1: Import the relevant libraries


Step 2: Read the dataset
Step 3: Load Pre-Trained VGG model
Step 4: Add Custom top layers
Step 5: Fine tune the model
Step 6. Evaluate the model
Step 7: Stop Programming

Theory

Transfer learning is an approach where we use a pre-trained model (trained on a large


dataset) as a starting point for a different task. Instead of training a model from scratch, we
reuse the learned features and adapt them to our specific problem. Suppose we have a pre-
trained model (e.g., a neural network) that has learned useful features from a large dataset
(e.g., ImageNet for image classification). We transfer the weights (learned parameters) from
this pre-trained model to a new model designed for a different task (e.g., classifying specific
objects in medical images). The main goal is to improve the performance on the new task
using the gained knowledge from the source task. VGG16 is a deep CNN architecture
developed by the Visual Geometry Group (VGG) at the University of Oxford. It has 16
weight layers, including 13 convolutional layers and 3 fully connected layers. The model was
trained on the ImageNet dataset, which contains millions of images across 1000 classes. It
gained prominence for its simplicity, uniform structure, and impressive performance on
large-scale image recognition tasks.

1
Program

import tensorflow as tf
from tensorflow.keras.applications import VGG16
from tensorflow.keras.layers import Dense, Flatten
from tensorflow.keras.models import Model
from tensorflow.keras.datasets import cifar10
from tensorflow.keras.utils import to_categorical

# Load CIFAR-10 dataset


(train_images, train_labels), (val_images, val_labels) = cifar10.load_data()

# Preprocess data (normalize pixel values, convert labels to one-hot encoding)


train_images = train_images / 255.0
val_images = val_images / 255.0
train_labels = to_categorical(train_labels, num_classes=10)
val_labels = to_categorical(val_labels, num_classes=10)

# Load pre-trained VGG16 model (excluding top layers)


base_model = VGG16(weights='imagenet', include_top=False, input_shape=(32, 32, 3))

# Add custom top layers


x = Flatten()(base_model.output)
x = Dense(256, activation='relu')(x)
output = Dense(10, activation='softmax')(x) # 10 classes for CIFAR-10

# Create the final model


model = Model(inputs=base_model.input, outputs=output)

# Compile the model


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

# Train the model


model.fit(train_images, train_labels, epochs=10, validation_data=(val_images, val_labels))

2
Inference:

Result:

You might also like