Prerequisite: Image Classifier using CNN
Image classification is a method to classify the images into their respective category classes using some method like :
- Training a small network from scratch
- Fine tuning the top layers of the model using VGG16
Let’s discuss how to train model from scratch and classify the data containing cars and planes.
Train Data : Train data contains the 200 images of each cars and planes i.e. total their are 400 images in the training dataset
Test Data : Test data contains 50 images of each cars and planes i.e. total their are 100 images in the test dataset
To download the complete dataset, click here.
Model Description: Before starting with the model firstly prepare the dataset and it’s arrangement. Look at the following image given below:
For feeding the dataset folders the should be made and provided into this format only. So now, Let’s begins with the model:
For training the model we don’t need a large high end machine and GPU’s, we can work with CPU’s also. Firstly, in given code include following libraries:
Every image in the dataset is of the size 224*224.
train_data_dir is the train dataset directory.
validation_data_dir is the directory for validation data.
nb_train_samples is the total number train samples.
nb_validation_samples is the total number of validation samples.
Checking format of Image:
This part is to check the data format i.e the RGB channel is coming first or last so, whatever it may be, model will check first and then input shape will be feeded accordingly.
About the following terms used above:
Conv2Dis the layer to convolve the image into multiple images
Activationis the activation function.
MaxPooling2Dis used to max pool the value from the given size matrix and same is used for the next 2 layers. then,
Flattenis used to flatten the dimensions of the image obtained after convolving it.
Denseis used to make this a fully connected model and is the hidden layer.
Dropoutis used to avoid overfitting on the dataset.
Denseis the output layer contains only one neuron which decide to which category image belongs.
Compile function is used here that involve use of loss, optimizers and metrics.here loss function used is
binary_crossentropy, optimizer used is
Now, the part of
dataGenerator comes into figure.In which we have used:
ImageDataGeneratorthat rescales the image, applies shear in some range, zooms the image and does horizontal flipping with the image. This ImageDataGenerator includes all possible orientation of the image.
train_datagen.flow_from_directoryis the function that is used to prepare data from the train_dataset directory
Target_sizespecifies the target size of the image.
test_datagen.flow_from_directoryis used to prepare test data for the model and all is similar as above.
fit_generatoris used to fit the data into the model made above, other factors used are
steps_per_epochstells us about the number of times the model will execute for the training data.
epochstells us the number of times model will be trained in forward and backward pass.
validation_datais used to feed the validation/test data into the model.
validation_stepsdenotes the number of validation/test samples.
At last we can also save the model.
Below is the complete implementation:
- keras.fit() and keras.fit_generator()
- Getting started with Scikit-image: image processing in Python
- Converting an image to ASCII image in Python
- Keras.Conv2D Class
- Building an Auto-Encoder using Keras
- Building a Generative Adversarial Network using Keras
- Python PIL | Image.tell()
- Python PIL | Image.getdata()
- Python PIL | Image.histogram()
- Python PIL | Image.new() method
- Image Captioning using Python
- Python PIL | Image.open() method
- Image Resizing using OpenCV | Python
- Python PIL | Image.transform() method
- Python | Crop image using pillow
If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to firstname.lastname@example.org. See your article appearing on the GeeksforGeeks main page and help other Geeks.
Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.