Scribd
Upload
100 views

ML Lab Session 06 - VGG16-CNN

The document discusses VGG16, a convolutional neural network architecture developed by researchers at the University of Oxford. VGG16 gained popularity for its simplicity and effectiveness in image classification tasks. It consists of 16 weight layers including convolutional layers to extract features from images, pooling layers to reduce information, fully connected layers for classification, and an output layer. The model was trained on the large-scale ImageNet dataset, containing over 1 million labeled images, and achieved high accuracy in image recognition. Pretraining VGG16 on ImageNet allows it to learn general visual features that provide benefits for other computer vision tasks requiring image classification.

Uploaded by

chatgptlogin2001
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
100 views

ML Lab Session 06 - VGG16-CNN

The document discusses VGG16, a convolutional neural network architecture developed by researchers at the University of Oxford. VGG16 gained popularity for its simplicity and effectiveness in image classification tasks. It consists of 16 weight layers including convolutional layers to extract features from images, pooling layers to reduce information, fully connected layers for classification, and an output layer. The model was trained on the large-scale ImageNet dataset, containing over 1 million labeled images, and achieved high accuracy in image recognition. Pretraining VGG16 on ImageNet allows it to learn general visual features that provide benefits for other computer vision tasks requiring image classification.

Uploaded by

chatgptlogin2001
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
You are on page 1/ 15

ITC2252 - Introduction

to Machine Learning
Practical Session - VGG16 in CNN
Steps of the process
01 Import Data
02 Clean the Data
03 Split the data to testing & Training
04 Design the model
05 Train the Model
06 Make Predictions
07 Evaluate and Improve
Today’s session

01 What is VGG16 Model?

02 Importance of ImageNet in VGG16


01
What is VGG16 Model?
Overview of VGG16 model & ImageNet dataset

Architecture & Layers Image Recognition

The VGG16 model is It has been trained on the


renowned for its depth, ImageNet dataset for image
consisting of 16 weight classification and has
layers, which makes it achieved state-of-the-art
suitable for complex image results in various
recognition tasks. competitions.
What is VGG16 model?
➔ The VGG16 model is a convolutional neural network architecture that was
developed by the Visual Geometry Group (VGG) at the University of Oxford.
➔ It gained popularity for its simplicity and effectiveness in image
classification tasks.
➔ The model basically consists Input Layers, 13 convolutional layers, 5 Pooling
Layers, 3 fully connected layers and Output Layer.
Layers in VGG16 model
1 Input Layers
The CNN receives input images and processes them through different layers, learning
features automatically from the data.

2 Convolutional Layers
These layers apply mathematical operations to the input data to extract high-level features
and patterns from the images.

3 Pooling Layers
Pooling layers condense the output of the convolutional layers, reducing the information to
the most essential elements.

4 Fully Connected Layers


Fully connected layers take the high-level features extracted by the convolutional layers and
perform classification.

5 Output Layer
The output layer produces the final classification probabilities using the softmax activation
function. (softmax activation function assigns probabilities to different classes)
Convolutional Block
VGG16 consists of multiple convolutional blocks, each composed of multiple
convolutional layers followed by max-pooling layers.
➢ Convolutional Layer
Applies convolutional filters to the input to
extract features.
➢ Rectified Linear Unit (ReLU)
Introduces non-linearity by applying the
ReLU activation function to the output of the
convolutional layer.
➢ Max-Pooling Layer
shrink the image size by keeping only the
most important information, making the
network efficient for image recognition
Fully Connected Layers Block
After the convolutional blocks, there are fully connected layers that take the high-level
features extracted by the convolutional layers and perform classification.

➢ Flatten Layer
Converts the 3D volume into a 1D vector.
➢ Fully Connected (Dense) Layer
Neural network layers where each neuron
is connected to every neuron in the
previous layer.
Importance of VGG16 model in CNN tasks

● One main importance of the VGG16 model is its strong performance in


image classification tasks.
● With its deep architecture and the ability to capture complex and
hierarchical features, VGG16 achieves high accuracy in recognizing and
categorizing images.
● This makes it a reliable and effective tool for various applications that
require accurate image classification, such as object recognition,
scene understanding, and medical image analysis.
02
Importance of ImageNet
in VGG16
What is ImageNet & ImageNet classification problem?

➔ ImageNet is a large-scale dataset that contains millions of labeled


images belonging to 1,000 different classes.
➔ The ImageNet classification problem refers to the task of classifying
images into one of several predefined categories or classes.
➔ The goal of the ImageNet classification challenge is to train a model
that can accurately classify images into the correct classes.
Importance of ImageNet in VGG16 model
➔ In VGG16 model (CNN Model), the pretrained weights are trained on the
ImageNet dataset (Pretrained weights refer to a set of learned
parameters that are obtained by training a model on a large dataset.).
➔ The reason why the ImageNet dataset is commonly used for pretraining
models like VGG16 is that it provides a diverse range of images from a
wide range of categories.
➔ By pretraining on ImageNet, the model learns general features and
representations that can be useful for various image recognition tasks.
➔ By utilizing the pretrained weights from ImageNet, the VGG16 model
already has knowledge of a wide range of visual features, such as
edges, textures, and shapes, which are beneficial for many computer
vision tasks.
➔ This pretrained initialization helps to speed up training and improve the
performance of the model on specific classification tasks, even if the
target dataset is different from ImageNet.
VISIT BELOW LINK TO GET THE CODE

Skin cancer detection example:


https://github.com/SanduNihara/VGG16_CNN.git
Thanks
Do you have any questions?

You might also like