walekarketaki8 / Flower_Classification_Using_Keras

Classify 5 types of flowers using Keras

Flower_Classification_Using_Keras

Aim : Classify 5 different types of flowers.
This project is done using Keras.Keras has ImageDataGenerator class which we used in this project allows the users to perform image augmentation on the fly in a very easy way.

Steps to follow to run the project:

1. Download dataset : To download dataset click here Dataset contains 5 folders, each folder represents different flower classes.

2. Directory Structure : Keras need directory structure of dataset to be in below format.
   
   To convert downloaded data into above format use Dataset Preparation Keras Format.ipynb which segregate data into test(20% of entire data) and train.
   

3. Training : Once train and test folders are ready, train model using command

python cnn.py

Here training is done with 30 epochs, for better accuracy customize parameters as per our requirements which includes input_shape, batch_size, nb_epoch. Also can modify CNN architecture by adding more layers and more nodes in each layer etc.

4. Testing : cnn.py. will give model.h5. To test on any image run below command

python test_cnn.py <image_path>

5. TRAIN_TEST_CNN.ipynb contains training and testing in single code. It will give results of model on test set in Pred.csv file.
   

6. Sample Result: