Image-Recognition-&-Classification-with-Keras-in-R-|-TensorFlow-for-Machine-Intelligence-by-Google

[fazelelham32]

Summary

The text outlines the process of using the EBIMAGE and Keras packages in R to create a simple image classification model that distinguishes between images of planes and cars. The process involves:

1. Installing and Loading Packages: EBIMAGE for image processing and Keras (with TensorFlow backend) for deep learning.
2. Image Collection and Saving: Saving six images each of planes and cars in a specified directory.
3. Setting Working Directory: Configuring the working directory to easily access the images.
4. Reading and Resizing Images: Reading 12 images into R and resizing them to a uniform 28x28x3 dimension.
5. Preparing Data: Converting images into vectors, splitting data into training and testing sets, and encoding labels (0 for planes, 1 for cars).
6. Building the Model: Creating a sequential neural network model with two hidden layers (256 and 128 neurons) using ReLU activation and softmax for the output layer.
7. Training and Evaluation: Training the model with 30 epochs, evaluating performance, and using a confusion matrix to assess accuracy.
8. Predictions and Probabilities: Generating predictions, calculating probabilities, and displaying misclassifications.

Highlight Keywords

• Packages: EBIMAGE, Keras, TensorFlow
• Images: planes, cars, saving, resizing, directory
• Data Preparation: training, testing, vectors, encoding
• Model: sequential neural network, hidden layers, ReLU, softmax
• Training: epochs, batch size, validation split
• Evaluation: confusion matrix, accuracy, probabilities, misclassifications