A quiz on the organization of the module

Adrian F. Clark

1. How many progress tests does the module have?

   • 1

   • 2

   • 0
   • 3

   There are two progress tests, roughly mid-way and at the end of the module.

2. How long is the module's exam?

   • 2 hours

   • 1.5 hours
   • 3 hours
   • 1 hour

   There is a single examination for the module or 2 hours duration, which will take place at the start of the summer term.

3. How many assignments does the module have?

   • 2

   • 0

   • 3
   • 1

   There are no assignments for the module.

4. What is examined in the progress tests?

   • the lectures and experiments

   • the experiments

   • the lecture material
   • your general knowledge of computer vision

   The progress tests focus on the experiments you have carried out, and that is why you are encouraged to keep detailed records of what you have done and what you have found. Lecture material is examined only where it is part of an experiment.

5. How is the module's material delivered?

   • a combination of printable notes, overheads and experiments for you to do

   • a series of videos with accompanying notes
   • lectures in which you make your own notes
   • laboratory sessions

   The primary resource for the module is the printable notes. I run through the material in a series of 'lectures' which gives you the chance to ask me questions -- though you can do that by email too. Vision is an inherently practical subject, so it is important that you also pick up practical skills by doing the experiments.

6. What is a restricted open-book exam?

   • you are provided with a copy of the leeture notes
   • you can access any web resources on your phone

   • you are given a summary sheet provided by the lecturer

   • you can bring in only a textbook

   For this module, you will be provided with a summary sheet of important stuff in the exam, and this will be available on Moodle before then for you to review. You will not be able to use an electronic version of them on a phone or similar. The idea is to make the exam a test of your ability to solve computer vision problems, not a test of your memory.

7. Why do I have to learn OpenCV?

   • it is really difficult to learn
   • it is really easy to learn

   • it has become the de facto standard for computer vision programming

   • it is the only computer vision package available

   OpenCV has become the most widely-used computer vision software package, especially for real-time work. It is quite ugly (in your lecturer's opinion) and quite difficult to use in places. Please do dedicate your career to improving upon it!

8. How are the progress tests organized?

   • on a one-to-one basis with the lecturer
   • online and without access to notes etc

   • online and open-book

   • in the software laboratories

   The progress tests are done online, with each student being given a random choice of questions from a 'question bank' and the order of the choices randomized, just like the test you are currently doing. Note that the progress tests are taken via Moodle and, unlike this one, you're allowed only one go at it. The tests are open-book, meaning that you can use your lecture notes and record of experiments in it.

9. What is the purpose of the lab sessions?

   • to learn OpenCV

   • to learn the practicalities and difficulties of vision software

   • to fill all my waking hours with work
   • to be able to do the progress tests

   I hope you realize that the real purpose of the experiments are to learn first-hand the difficulties of producing working vision software and to gain an understanding of what is possible in practice, even if it seems like you are doing the other three possible answers!

10. Why do the labs have to be done in Python?

    • it is good for rapid prototyping and development

    • your lecturer doesn't know Java
    • it makes using a test harness easier
    • it runs more quickly than other possible languages

    The real reason for having you program in Python is that it is currently the best language for rapid prototyping and has been interfaced to a wide range of external packages, including OpenCV and TensorFlow. Java is widely used but its notion of "write once, run anywhere" is slowly being eroded and its run-time support is often ugly. The Java wrappers for OpenCV also have a reputation for being difficult to use. Programming OpenCV applications directly in C++ is worthwhile only when there is a real need for speed. On the other hand, Python code can execute pretty slowly.