Multi-Class-Classification-By-using-Keras-Deep-Learning-

IMPORTANT INSTRUCTION

run code in this order

1. sgsp_data.py #it will maintain datasets
2. sgsp_model.py # it will maintain model
3. sgsp_main.py # it will predict on visualise model

Design

In this section I wants to predict the class : here first i click photos of my friends from image_detect.py code

I take 4 classes ["gopal","pradeep","sannidhya","shivam"]. But you can take multiple classes . steps:

1: After capture images of my friends . i divided images into training_data and testing data

2: Then resize the images eg:(50*50)

3: convert it into numpy array

4: reshapr numpy image

5: convert training_data into x_train(features),y_train(labels)

6: then convert y_train into categorical (By using OneHotEncoder)

eg: if you have 2 classes

               0  1
               0  0
               0  0
               0  0
               1  0
               0  0
               0  0
               
              
              
              only one row take 1 other remains 0 .thats why its called OneHotEncoder

7: then take model

8: make layers .its depends upon you ,how much layers you wants to take.

9: compile the program (optimizer,loss,accuracy)

10: training our model (model.fit . or model.fitGenerator)

11: predict on test data

12: visualize on graph