One of the best ways to engage with new material in computer science is to implement it yourself. Accordingly, over the course of the semester you are expected to develop a portfolio of work showing the exploration and implementation you have carried out as you explore the course material. You are encouraged to tailor the contents of this portfolio according to your interests as they develop.
Although the final portfolio review will take place at the end of the semester, periodic checkpoints have been established to assist you with time management. At each checkpoint you must submit at least two new items for your portfolio, and you may optionally revise old ones as well by submitting an updated version. By the end of the semester, your work will be assessed for both breadth and quality.
Here is the list of checkpoints, with minimal assignments to be included. Throughout the semester there may be additional writings and other pieces that you will add as well.
- Week 2
- Literature review & one exercise
- Week 5
- Three additional exercises
- Week 8
- Two additional exercises
- Week 10
- Two additional exercises
- Week 12
- Final project complete
- Week 13
- Final portfolio revisions due
By the end your portfolio should contain at least eight pieces of independent work (mostly exploratory exercises and the a literature review), plus the slides for your two presentations and the writeup of the final project. Other potential portfolio items of independent work might include significant teaching units developed for the class, and reports written after attending a research talk about some current computer vision topic.
An essential piece of the portfolio process is the opportunity to revise previously submitted work based upon feedback. For each submission, I will provide feedback and commentary about what you have done, and identify suggested areas for improvement (if any). If an exercise does not fully meet expectations, you are encouraged to submit a revised version with a later checkpoint. The final version of the item is the one that forms the basis of your grade.
Format
Please submit each exercise as a single document, either .PDF or .DOCX. (If you use Jupyter notebooks, please submit both the raw notebook and a PDF transcript.) The document should contain your written narrative describing what you did, how it went, and what you learned, as well as code (can be pasted from your program) and images (screenshots or saved files) also pasted in. Please make sure that the narrative focuses on what you have learned and understood from the exercise rather than just presenting results with no commentary.
With each checkpoint submission you will include a manifest (table of contents) listing each item you are including, whether new or revised. The listing should include a revision history describing when the item was first submitted and any subsequent revisions. In particular, it should clearly identify all material that is either new or revised in the current submission -- this is important so that I don't waste time grading material that I've already seen. A missing or incomplete manifest is grounds for a lower grade.
Resources
One approach is to find a tutorial on a particular technique and work through it for yourself. If the tutorial gives you code and does not involve writing your own, then you should take the code as a starting point and try some sort of modification or extension to learn how it works. In all cases you should cite the source of the material, and the piece that you turn in should include enough narrative (either as a separate document or comments interspersed with code) that I can follow what you are doing, and what piece is your own.
There is a lot of computer vision course material on the web, so you can find many exercises by searching. You are encouraged to find activities on topics that interest you through experimentation and web research. If you have not taken CSC 268 you can choose to include assignments and/or beefed up lab exercises (extended by you) for some of your portfolio elements. You can also look at the list of exercises that were used in lng-ago versions of this course. Caution though -- some of these are quite out of date. Try not to rely exclusively on these listings -- be exploratory!
Final Portfolio Submission & Assessment
Your final portfolio submission should collect all your work over the course of the semester into a single package. It should contain at least eight items, plus a literature review. Additional items may be included at your discretion. Previously submitted items may be revised for the final submission. You should try to revise all items where the initial comments indicated any shortcomings.
Portfolios will be assessed based on the quality, originality, and extent of the final submission. I am expecting competent completion of each chosen exercise, with a clear presentation of the results. An A portfolio should include several examples of work that exceeds the norm, either through creative exploration of a technique that goes beyond beyond the basic material, thorough and clear analysis that elucidates a fine point of the results, or other similarly exceptional work.