Project3-ImageClassification-ConcreteCrack

To classify the image of concrete wheteher it is crack or not link: https://data.mendeley.com/datasets/5y9wdsg2zt/2

1. Summary

To make classify the image of concrete
to predict the new image of concrete
deal with a large dataset
the deep learning model is used and trained
The model use Transfer Learning from mobilenetV2

2. IDE and Framework

The project built with Spyder as the main IDE
use Tensorflow, Keras, Numpy, Mathplot

3. Methodology

The folder contain 2 type of image which is positive(the concrete is crack) and negative ( the concrete is not crack).
There are 40 000 images, 20 000 in positive folder and 20 000 in negative folder.
The we classify the images into training and validation dataset. 70% of total images used for training and 30% of images for validation test. 12 000 use for validation and 28 000 for training
we create a pipeline for data augmentation including random flip and random rotation, to increate data's varities and prevent verfit in data training.

Model

use base model from mobilenetV2, uninclude the top layer ( generally for classification task) and freeze the trainable layer
Set the base model layer below 100 layer into false, so that the model will not update its weight / paramter during training;

Create a classification layer as top layer of the base model. Use global Average Pooling 2D, activation function of 'softmax' and number of class is 2 because we want to classify the image of concrete into 2; positive or negative.
Use Functional APi in creating Transfer Learning model:

Now, we need to train the model to update the trainable parameters.

Model Evaluate

-The model is compile with optimizer of 'adam' with learning rate = 0.001, loss= Sparse Crossentropy', metrics of accuracy, batch_size of 32 and epochs of 200

The value is display by using TensorBoard:

Model Prediction

Predicting a new image