Ad
More Related Content
What's hot (20)
Similar to Image classification using cnn (20)
Ad
More from SumeraHangi (7)
Ad
Recently uploaded (20)
Image classification using cnn
- 1. Image Classification using CNN Deep Learning Created by Sumera Hangi Roll no =06
- 2. CONTENTS – INTRODUCTION (core subject) – INTRODUCTION TO CONVOLUTIONAL NEURAL NETWORKS – WHY DID WE USE A CNN? – STEPS TO DESIGN CNN – KEY DIFFERENCE BETWEEN ANN & CNN – ARCHITECTURE OF A CNN FROM INPUT IMAGE TO IMAGE REGONITION – APPLICATIONS OF CNN – CONCLUSION
- 3. INTRODUCTION (core subject) – IMAGE PROCESSING is a method to perform some operation on image, in order to get an enhanced image or extract some useful information from it. – COMPUTER VISION is a field of computer science that works on enabling computer to see, identify & process images in the same way that human vision does, and then provide appropriate output. – IMAGE CLASSIFICATION is a complex process that may affected by many factors. – The objective of image classification is to identify and portray, as a unique gray level(or color), the feature occurring in an image in terms of the object or type of land cover these feature actually represent on the ground . – Image classification is perhaps the most important part of digital image analysis.
- 4. INTRODUCTION TO CONVOLUTIONAL NEURAL NETWORKS – CNNs are basically combination of convolutions followed by feature maps, subsampling, image features, dense layer of neural network . – We are using convolutions ,subsampling , feature maps to extract image feature & dense layer of NN to categorize input image according to image features. – Applications of CNN are image & video recognition, object recognition, natural language processing.
- 6. SO, WHY DID WE USE A CNN? – In machine learning, a convolutional neural network is a class of deep, feed- forward artificial neural networks that has successfully been applied to analyzing visual imagery. – CNNs use a variation of multilayer perceptron's designed to require minimal preprocessing. They are also known as shift invariant or space invariant artificial neural networks (SIANN), based on their shared-weights architecture and translation invariance characteristics.
- 7. STEPS TO DESIGN CNN – Convolution – Max pooling – Flattening – Full Connection
- 8. CONVOLUTION – A convolution sweeps the window through images then calculates its input and filter dot product pixel values. This allows convolution to emphasize the relevant features.
- 10. IMAGE 3 IMAGE 4
- 11. RELU ReLu is a non-linear activation function that is used in multi- layer neural networks or deep neural networks. This function can be represented as: where x = an input value According to equation 1, the output of ReLu is the maximum value between zero and the input value. We are using Relu on the convolution to increase non linearity
- 12. MAX POOLING CNN uses max pooling to replace output with a max summary to reduce data size and processing time. This allows you to determine features that produce the highest impact and reduces the size of the image by 75% to avoid the risk of overfitting also get image extracted features.
- 13. FLATTENING Flattening the matrix to get a single column image based features
- 14. FULL CONNECTION – Neurons of hidden neural network layers are fully connected with all the features – Activation function determines the value of the output layer Softmax • We use softmax activation function in the output layer of multiclass CNN system. • It uses multiple classification logistic regression model. • It calculates the probability distribution of the event over all the different events. • The range of the output of softmax is 0 to 1 & the sum of all the probabilities will be equal to one
- 15. FULL CONNECTION Single column fatten data works as input layer of fully connected ANN
- 16. KEY DIFFERENCE BETWEEN ANN & CNN – In CNN, neuron of hidden neural network are fully connected . – In CNN , input feature are connected with all the neurons. But , in ANN , input features are connected with only those neurons that are having similar values or pattern.
- 17. ARCHITECTURE OF A CNN FROM INPUT IMAGE TO IMAGE REGONITION
- 18. APPLICATIONS OF CNN A)COMPUTER VISION1) Face Recognition: 2) Scene Labelling: 3) Image Classification: 4) Action Recognition :
- 19. CONTD.. 5) Human Pose Estimation 6) Document Analysis:
- 20. B. NATURAL LANGUAGE PROCESSING 1) Speech Recognition 2) Text Classification
- 21. CONCLUSION – Convolutional neural networks (CNNs) have accomplished astonishing achievements across a variety of domains, including medical research, and an increasing interest has emerged in radiology. – Although deep learning has become a dominant method in a variety of complex tasks such as image classification and object detection, it is not a panacea. – Being familiar with key concepts and advantages of CNN as well as limitations of deep learning is essential in order to leverage it in radiology research with the goal of improving radiologist performance and, eventually, patient care.