Consider images of the following food items:

1. Donut

2. Onion Rings

3. Pizza

4. Spring Rolls

5. Steak

The images are grouped in three directories for training, validation, and testing. The training directory has 700 images for each category, the validation directory has 200 images of each category, and the test directory has 100 images of each category.

Use TensorFlow with Keras to build and train a Convolutional Neural Network (CNN) that can recognize each food category.

Please note that the images have different sizes and you will need to resize them to make sure they have the same size before using them in your model.

Once your model is trained, you will need to evaluate the model and provide the following:

1- The graph showing the loss during training.

2- The classification report.

3- The Confusion Matrix.