Object classification using deep neural network
- 1. Cat classification Presentation by- Nisha Kushwah Avani Mishra Sneha Tiwari Nikita Tomar
- 2. Contents- ● Basic understanding of “How Computer Understands Images” ● How neural network works. ● Neuron in the network ● Training/Test set ● Flow chart of object classification
- 3. How Computers Understand Images The smallest element of an image is called a pixel, It’s basically a dot in the picture. But a computer does not understand pixels as dots of color. It only understands numbers. (0,0,0) (135,206,235) (0, 0, 255) (225,0,0) R G B (pixel values)
- 4. Every image has three layers red, green and blue.
- 5. How neural network works In neural network an input layer gets the data into the “hidden layers” and after mathematical computation we can see the information provided by the output layer.
- 7. In the second layer, input pixels group together and make small parts of 9 digit picture represented as white nodes, in the next layer these small parts group again and make bigger parts of 9 digit image which give the output as 9 in the next output layer.
- 8. Example
- 9. Neuron in the Network Weights(W)—A weight represent the strength of the connection between units. If the weight from node 1 to node 2 has greater magnitude, it means that neuron 1 has greater influence over neuron 2. A weight brings down the importance of the input value.
- 10. Weights(w) update continually by getting feedback from the error calculated at every iteration, we calculate error many times and change the values of w to minimize error. Error reduces with every iteration.
- 11. Training and Test set ● First we train our model with training images, Then we test images from the trained model. ● We have 209 training images and 50 test images.
- 12. Flow chart of Object classification using deep neural network. Relu: Relu used to get intermediate numerical value, which represent the small group of pixels(or small part of image). Relu gives positive value between 0 - 1. Sigmoid: Sigmoid function used to get the binary value 0 or 1. We use it only for output to predict the output 0 or 1.
- 13. Thank you