MobileNetV2 transfer learning

For the assignment, the data are divided into two categories. The training set and the validation set. Will the validation set also be seen by the model to improve the model?
For real world scenarios, do we have to add a test set to see the real performance?

The validation dataset will be used to improve hypermarameters to fit the dataset as best possible. Yes, the test split is also advisable and very practical to use because ultimately it wont have any interaction with the model at all.