Character Recognition Using Neural Netwoks
-
Upload
nirav-kothari -
Category
Documents
-
view
229 -
download
0
Transcript of Character Recognition Using Neural Netwoks
-
7/31/2019 Character Recognition Using Neural Netwoks
1/12
-
7/31/2019 Character Recognition Using Neural Netwoks
2/12
Kaushik Patel
Nirav Kothari Roshni Gehlot
Hetal Kachhi
-
7/31/2019 Character Recognition Using Neural Netwoks
3/12
Artificial Intelligence is the science of making intelligent machines,especially intelligent computer programs, what is related to the processof understanding human intelligence with the help of computers, butArtificial Intelligence is not restricted to biologically observable methods.
In the course of 50 years of research, AI has developed a large number oftools to solve the most difficult problems in computer science. A few ofthe most general of these methods are discussed below: -
1. Search and Optimization.2. Logic.
3. Probabilistic methods for uncertain reasoning.4. Classifiers and statistical learning methods.5. Neural networks.6. Languages.
-
7/31/2019 Character Recognition Using Neural Netwoks
4/12
Conversational Behavior
Data Mining
Driverless Cars
Robot Soccer
-
7/31/2019 Character Recognition Using Neural Netwoks
5/12
Network of interconnecting artificial neurons.
Neurons are basically, decision-taking units.
Artificial neural networks may either be used to gain an understanding of
biological neural networks, or for solving artificial intelligence problems
without necessarily creating a model of a real biological system.
Computers are endowed with humanlike abilities to take decisions based
on current scenarios and circumstances.
-
7/31/2019 Character Recognition Using Neural Netwoks
6/12
Feedforward Neural Network
Radial Basis Function Network
Kohonen Self-Organizing Network
Learning Vector Quantization
Recurrent Neural Network
-
7/31/2019 Character Recognition Using Neural Netwoks
7/12
Operates in two modes, namely: -
1. Training &
2. Mapping
Comprises of Nodes or Neurons.
Every node is associated with its respective weight vector of the same
dimension as the input data vectors, and a position in the map space.
Forms a semantic map where similar samples are mapped close
together and dissimilar ones apart.
No Hidden Layers.
-
7/31/2019 Character Recognition Using Neural Netwoks
8/12
It is simple and gets trained rapidly.
There are no hidden layers.
Output comprises of only one neuron. Thus only one winning neuron is
selected that will be closest to the input sample.
-
7/31/2019 Character Recognition Using Neural Netwoks
9/12
-
7/31/2019 Character Recognition Using Neural Netwoks
10/12
1. Draw the Image and Enter corresponding character in text box: -
Use the mouse as an input to draw the image of characters. Enter the
corresponding character with which you want the drawn character to be
recognized in the text box.
2. Down sampling of Image: -
The images have to be cropped sharp to the border of the character in
order to standardize the images. The image standardization is done by
finding the maximum row and column with 1s and with the peak point,
increase and decrease the counter until meeting the white space, or the
line with all 0s. This technique is shown in figure below where a character
S is being cropped and resized.
-
7/31/2019 Character Recognition Using Neural Netwoks
11/12
The image pre-processing is then
followed by the image resize again
to meet the network input
requirement, 5 by 7 matrices,
where the value of 1 will be
assigned to all pixel where all 10 by
10 box are filled with 1s, as shown:
Finally, the 5 by 7 matrices is
concatenated into a stream so
that it can be feed into network of
35 input neurons.
-
7/31/2019 Character Recognition Using Neural Netwoks
12/12
3. Training the Neural Network: -
In training the neural network we actually make the neural network
learn how to recognize image. The process of training is adjusting the
individual weights between each of the individual neurons until we
reach close to the desired output.
4. Recognition: -
Draw the image to be recognized in drawing area and click on
recognize button. The best neuron will be fired as output and
recognition will be performed.