Long-Range Planning
CSC 370 embraces the philosophy of "high control, high value" (Cavanagh, 2016). This can be intimidating, but the potential rewards are huge. When you take responsibility for your own educational experience, you can explore the field in the way that provides you with the greatest possible meaning. You also lay the foundation for a lifetime of self-directed learning.
For the first week your task is to map out a proposed course of study over the rest of the semester. Each student will prepare their own proposal to bring to class on the second day. We will then develop a shared plan as a group based on these individual contributions.
Computer vision encompasses a large variety of complex topics, more than will fit into a single semester course. To develop your personal proposal (which will become a part of your portfolio) you will have to pick and choose among different options -- which means that you will have to quickly develop a rudimentary knowledge of what each topic entails, and how they fit together. Which topics are foundational material that will be necessary for understanding everything else? Which topics stand alone, and may be considered in any order? Which do you find most fascinating and compelling?
To prepare for this assignment you should begin by reading through the introductions and contents to the course textbook(s). You can also take a look at the first section of each chapter for more detail on those covering topics that seem completely unfamiliar or unknown to you. Finally, you should locate a computer vision course taught at another institution and compare the topics on its syllabus with your own list.
If you have time, it is also worth browsing the topics addressed in computer vision conferences (listing is slightly out of date, unfortunately, but you can search for more recent versions of the main ones) -- some topics, such as deep learning for computer vision, are new enough that they are not well covered in the texts. Next, consider the following questions (you might make a set of lists):
- Which topics interest you the most?
- Which seem to have attracted the most attention from the research community? (There's probably a reason...)
- Which topics support the ones you listed above, either as a preliminary step or as a conceptual basis?
- Which topics seem foundational, such that they are necessary for a broad understanding of the field?
Once you have made these lists, use them to pick and choose a set of topics that can be considered over the eleven available weeks of the semester. (We will use the first week for planning, and the last week for presentations of final projects.) Having chosen the topics, you will also need to decide on an order. You probably should not introduce more than one new topic per week (unless a group of closely related topics can be considered together). On the other hand, particularly involved or interesting topics may be given more than one week -- deep learning methods will probably take significant time. In such cases, you should try to break the treatment down into a weekly set of subtopics.
Weekly Planning
Following the first week, planning will take place on a weekly basis, with different teams of 3 responsible for each week. Each student will help to plan at least two weeks of the semester.
In a normal week, one session could be devoted to presentation and discussion of research papers and the other to background studies or activities that relate to the topic of the week. This could take the form of a lecture, a lab activity, or some group exploration. The presentations do not need to be made by the same students who are planning the week, although that is usually the simplest way to arrange things.
The week prior, each planning team should meet with the professor to go over the upcoming plan. (Most likely these meetings will be on Fridays during the course time block.) Before the meeting, you should undertake a literature review of the area in question. This should prepare you to answer the following questions:
- What problem or problems are being solved?
- What are the main approaches, historical and current?
- What are some (at least 3-5) key papers that made significant contributions?
- What obstacles and drawbacks currently remain?
Having answered these questions for yourselves, you should then propose a set of activities for the group. These should normally include relevant academic papers to be presented and discussed, together with the groundwork for understanding them. The plan may include a professor lecture on content as one component, but should also include time for discussion and an area of activity.
All students on a team are responsible for planning the week. However, each will have a specific job to help focus the duties.
- Facilitator
- Each week will have one or two facilitators, who are responsible for carrying out the group's plan during the two classroom periods. They will lead the discussion, transition from one topic to the next, etc. Each student must sign up to facilitate during one of their two weeks.
- Communicator
- The person in this role is responsible for helping the rest of the class to prepare for the seminar meetings. By midnight of the preceding Friday, they will post to Moodle a planned syllabus for the week and email the group. In particular, this should include any papers that will be presented during the week so that all students will have a chance to read them in advance.
- Recorder
- This job takes place during and after the class meetings. The recorder's job is to take detailed notes of the class discussions and activities, and post a record of these to Moodle. To the extent possible, these should draw links to material covered in earlier weeks as well.
Teams should decide on these roles before the planning meeting with the professor. Work on planning teams will form a significant part of the participation grade for the course.