Using Neural Networks Through Tensorflow: Image Recognition

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 14

IMAGE RECOGNITION

using Neural Networks


through TensorFlow
Case studies of Image recogniton and TensorFlow :
Agenda:
• Neural Networks
• TensorFlow
• Image Recognition Through Dense Neural Networks
using TensorFlow
Neural Networks
• Neural Networks is an application of ML, which works
similar to human brain.
• Neural Networks take Data as an input, train themselves
to recognise the patterns in the data, and try to predict
the output.
TensorFlow:
• It is a end-to-end open source platform used to
implement Machine Learning Algorithms, particularly
Neural Networks.
• Developed by Google brain team, initially released in
2015.
Tensors:
• A Tensor is an algebraic object that describes a
(multilinear) relationship between sets of
algebraic objects related to a vector space.

• For an informal defenition,


Tensors are generalisation of Vectors.
EXAMPLES of different Tensors:
Why TensorFlow?
1. Versatility: Can be run on a Desktop, MobileDevices
or even on cloud.
2. Portability: Train on one machine, Run on any other
machine also.
3. Speed: Developed in C++, can be accessed by other
high level languages like python.
4. Community: It is highly popular and has the biggest
community compared to CNTK, Theano and Pytorch.
Image Recognition
Fashion MNIST data set from Keras:
Sample Images from Fashion MNIST data set:
1. Data Collection / Data Set Collection
2. Data Pre-Processing
3. Building the Model
4. Compiling the Model
5. Training the Model
6. Evaluating the Model
7. Predicting the Model
8. (*optional)Verifying the predictions

You might also like