Course on Classical and Modern Papers in Image Analysis
Once a month, according to the per-session selected 2-hour time-slot (but aiming for a regular slot every 2nd Monday in a month, 15:15-17:00), we discuss a paper within any of the interest fields of the PhD students at the Computerised Image Processing doctoral programme. We aim at (each month) alternating between a classical and a modern paper (from the last 5 years or so, but established enough to have generated sufficient attention from the community). The lists of "articles that have stood the test of time" in the field of computer vision and pattern recognition and in the field of artificial intelligence could provide inspiration.
See here for the schedule and selected papers since 2011.
Each session, one PhD student acts as a discussion leader. The discussion leader is expected to give an overview of the paper; it is up to the discussion leader to decide what is a suitable level of detail for this introductory presentation. Everybody is expected to read the paper, and to submit a commentary or discussion questions to the discussion leader (usually via a shared document prepared by the session leader). The commentary can be about anything in the paper: positive or negative criticism about the paper as a whole, or about something that is poorly explained, an interesting detail, a surprising solution, etc. During the session, these points are discussed in further detail, along with other points that the discussion leader finds interesting.
0.3 ECTS are awarded to all the students who are actively present and who submitted a discussion question/commentary. 1 ECTS is awarded to the discussion leader. The accumulated credits for the whole course are registered when requested (typically close to a thesis defence time), and only if 3 or more ECTS have been collected. Other researchers or students are welcome to participate in the seminars (but are typically not given course credits, unless special arrangements are made).
The course participants are invited to join a discussion group (we currently use RingCentral a.k.a. Glip for that purpose). All the internal communication, in particular regarding organization of the sessions (selection of papers, scheduling the seminars), is kept there.
The discussion leader is responsible for all aspects of the organization of the discussion session. Please, use this checklist when preparing for the session:
- Choose a paper together with your supervisor and share it with the course participants (share a pdf of the paper or a link to it). Prepare a shared document which can be used by the participants to submit their questions and commentaries. Start well on time! Aim to allow at least one week for everyone to read the paper to be discussed.
- If the default time-slot is not suitable: Schedule the session (by initiating a Doodle), aiming to enable participation of as many course participants as possible, as well as participation of your supervisor. Note that the course coordinator's (currently Joakim) presence is not mandatory (but he aims at joining whenever possible).
- Room 104150 is booked for the sessions. If any changes appear: the discussion leader should book a suitable seminar room (via TimeEdit), or schedule an online session (via Zoom). Share this information with the course participants.
- The sessions are included in the Vi3 calendar. If any changes of the default time and place are made, the discussion leader should send an email to the it-vi3 (at) lists.uu.se with the information about the time and place of the session (once it is decided among the course participants) and the paper to be discussed (attaching a pdf to the mail).
- Update the schedule page with the relevant information. (You can update the page by clicking on the "Edit this page" link in the rightmost column in the black bar at the bottom of the page.)
- Study the paper carefully and prepare to present it (at a suitable level of detail) to all who joined the session. Note that everybody is expected to study the paper, not just the discussion leader! This can help the discussion leader to focus on more interesting aspects of the paper, rather than explaining all the details. It is often informative to not only look at the selected paper, but also at earlier and/or later significant papers that are strongly related.
- Collect the questions and commentaries that other students submitted, and organise them in a suitable way on your slides. These should be the basis for a discussion of the paper. A suggestion is to include them at different suitable points in your presentation, rather than addressing them all at the end.
- Be prepared to provide (print or send electronically, depending on the form of the session) your supervisor with this form. Do this before the start of the session. Mark the names of the students who submitted a commentary. Your supervisor is responsible for indicating activity of the participants (leading to assigned credits). The filled-in form should be given to Nataša. When present Nataša takes care of this step herself, and no action from the supervisor is needed on this matter.
- Try to get a discussion going! It is important to prepare a good presentation, but it is at least equally important to activate others. Note that it is not your duty, as the session organiser, to answer all the questions posed by the participants! It is rather that the participants should always aim at indicating interesting questions that they are happy to discuss further.
- For simplicity, there is a suggested annual schedule indicating the session organiser for each month of the academic year - this list is agreed on by all the participants at an early point to help everyone plan their time. If you cannot organise the session according to this schedule, then it is your responsibility to be proactive and find a replacement. You can swap with any course participant who did not already present during the ongoing academic year.
Possible topics for commentary and discussion:
- What were the main pre-existing methods that the authors based their work on?
- What later methods have been based on these results?
- What have the authors accomplished with their results?
- What have others accomplished with these results?
- Have other people come up with the same ideas?
- What would our field look like today without this paper?
- Scientific content
- How this content is presented
- How the methods are described
- How the methods are evaluated
- How the results are described
- How the results are discussed
- Writing style
- Does the abstract correctly represent the contents?
- What is the reader assumed to know?
- Overstatement/understatement of results