Image Based Recognition and Classification: Assessment
Assessed Tasks
There will be three separate parts to the assessment of this
course. The first part will not be graded but will have to completed
to pass the course.
Homeworks Pass/Fail
At the end
of each lecture a relatively simple programming task and/or pen and
pencil exercise will be assigned. These will have to be
attempted/finished and brought to the next lecture, where they will be
corrected by a fellow student. Successful completion of
take-home assignment will be defined as:
- coming to the lecture with the assignment attempted/completed
(determined by your peers)
AND
- correcting a fellow student's assignment (according to a sample solution)
There will be 11 such assignments and to pass you must successfully
complete 9 of the tasks.
Lab Project Pass/Fail
In groups of one to three you will implement a face detector which
is learned using the boosting algorithm. This part of the course
will be assessed as follows:
Lab Part I
-
Successfully complete set tasks given default data and settings
used for debugging.
- Compute ROC curves given test data using
your final classifier.
- Run your final classifier face
detector on the test images in which the face(s) are only a small
portion of the image.
NOTE: Lab Part I
must be completed to pass the lab project and the course.
Lab Part II Bonus points available (upto 10%
added to final exam score)
Make a poster presentation, one from each group, about your face
detector. These will be marked by a combination of the course
leader, teaching assistants and a subset of fellow students at
an afternoon session where all the students will present their
posters and visit and grade other posters. To obtain high marks
you have to have a clear poster and be able to comment on and
discuss issues such as
- design choices with respect
to the weak classifier,
- computational issues with learning a
boosting classifier,
- the performance of your final detector
on a set of test images.
Final Exam Final grade
Your final grade for the course will be determined by final
exam. There will be a complusory part of the exam Part I,
covering core ILOs, those in the Part I ILOs list, which
will have to be answered at a high level to pass the exam
(E). While to obtain higher grades other questions,
covering Part II ILOs and Part I ILOs at a higher
level of understanding, will have to be answered.
Grading
To pass the course, ( = grade E), the following criteria
have to be fulfilled:
- Completion of 9 homework assignments
- Completion of Lab Part I.
- Score at least a 70% on Part I on the final exam.
To get a higher grade
- Perform well on Part II of the final exam and/or
- Participate and perform well on the poster presentation.
Credits
The credit associated with each assigned task.
Homeworks 1.5 hp
Lab Project 2 hp
Final exam 2.5 hp