https://www.kaggle.com/chrisfilo/fruit-recognition#Apple%20102.png
Dear students,
I have received a lot of reports to review from students, some of them have been excellent, which is good to see. From the feedback I have given, I thought I would summarise it for all students:
* Make sure you are using a dataset of images, at least 5000 images with at least 3 classes
* Use a process of either 1. image processing with neural network, OR 2. deep learning neural network. Any other technique is not correct, e.g. SVM, etc.
* Split your data properly, 70% for training, 10% for validation, 20% for testing.
* You need to include accuracy (%) results for all, training, testing and validation.
* The training curves should include the training, testing and validation result for each epoch
* Make sure the training continues until the results plateau
* The confusion matrix should be used to visualise the testing results
* Use the template included to structure your report.
* Double-check the coursework description and marking scheme to make sure you have included everything that is required.
Good luck!