COM SCI 163
Deep Learning for Computer Vision
Lecture and discussion. The registrar has published a final exam time.
Fall 2026: 1 lecture and 2 discussions. All of them are open. The oldest of these readings was 79 minutes ago.
Grades
3.36
Average grade awarded in COM SCI 163. Not a student rating: this is what the grades were.
In the A range
49%
of the 255 letter grades
D, F or W
1.2%
of every letter grade awarded, plus withdrawals
Taken pass/no pass
-
the registrar has this letter grade only this term
The registrar's own records: 255 grades over 2 terms, Fall 2021 to Fall 2025. Every instructor is pooled here.
What is inside that record
No summer session is in that record.
The table below splits them, and term averages across those 2 terms ran 3.31 to 3.41.
Historical grades for ZHOU, BOLEI, across recorded terms. The course summary above includes all professors.
- A+ 20 · 7.8%
- A 46 · 18.0%
- A- 59 · 23.1%
- B+ 50 · 19.6%
- B 37 · 14.5%
- B- 24 · 9.4%
- C+ 8 · 3.1%
- C 5 · 2.0%
- C- 3 · 1.2%
- D+ 1 · 0.4%
- F 2 · 0.8%
Percentages are of the 255 letter grades. Grey bars are non-letter outcomes.
How the non-letter outcomes are counted
They are excluded from the GPA entirely rather than scored, because counting a P as a 4.0 would be a fabrication.
By instructor
| Instructor | Predicted GPA | Raw | n | Terms | A range | Workload | Fall 2026 |
|---|---|---|---|---|---|---|---|
| ZHOU, BOLEI reviews ↗ | 3.36 | 3.36 | 255 | 2 | 49% | average | Open, 11 seats left |
Why Predicted is not the Raw average
“Predicted” is shrunk toward the course average, which is itself shrunk toward the department, so an instructor with a handful of students sits near the course mean rather than topping the list.
Grade records describe outcomes. For what a course is like to take, read student reviews written on dibs below or follow the separate external review link.
Workload and assignments
Work types reported in BruinWalk reviews of this course, across recorded instructors and terms. They may differ from this term's syllabus.
dibs has not read this course’s reviews, so it cannot say what kind of work it sets. That is dibs not having looked. It is not a course without papers or exams, and the reviews linked below are the answer until dibs catches up.
What comes first
The chain dibs can see is 7 courses long before COM SCI 163. This is a floor, not a deadline: dibs has not read a class page for COM SCI 145, COM SCI 162, EC ENGR 10, EC ENGR 149, EC ENGR C147, MATH 170A, and MATH 3B, so the real chain can only be longer than this, never shorter.
Counted in courses rather than quarters: dibs does not know which of them a department lets you take together.
From the registrar's catalog
Lecture, four hours; discussion, two hours; outside study, six hours. Requisite: one course from course C124, 145, M146, M148, 161, 162, Electrical and Computer Engineering C147, or 149. Computer vision has been a core field of artificial intelligence, facilitating a wide range of applications from image search to self-driving. The progress of deep learning has greatly advanced the performance of visual tasks like visual recognition and image generation. Study of deep learning approaches for computer vision. Students learn to implement and tune the deep neural networks used in various computer vision, such as visual recognition and image generation. Covers learning algorithms, neural architecture design, and practical skills of training and debugging neural networks. Letter grading.
Sections in Fall 2026
You enroll in the discussion, not the lecture, so a discussion's status is the one that decides whether you get in. A lecture can read open while every discussion under it is full.
Seat counts as of : the oldest of these 3 readings. Seat readings are snapshots, not live availability. Their timestamps show when dibs last checked.
-
Lec 1 MW 4:00pm-5:50pmOpen3.36predicted, this instructor · 255 grades
11 seats left. 109 of 120 enrolled high risk
2 discussions, all open.
This limit is on the class, so it applies to all 2 discussions below.
Final enrollment, averaged across each term's sections: between 117 and 138 a term over 2 terms, summer excluded, most recently 117 in Fall 2025. That is the shape across years, not this term, which is above.
From the people who took it
Student reviews
What taking COM SCI 163 was like: the work, the teaching, and what students wish they’d known.
Your experience could be the first.
No student reviews are available to read on dibs for this course yet. Took it? Tell the next person what helped, what was hard, and how you spent your time.
Looking for more perspectives?
External reviews are separate and do not count toward the dibs rating.
Seats over time in Fall 2026
How each lecture has filled since dibs first saw it.
A dot is a reading where something moved: the seat count, the status, or the waitlist.
What a flat run means, and where a line starts
Flat between dots means it did not drift, it held and then jumped, and a line starts when dibs first saw that lecture rather than reaching back before it existed. A lecture that has never moved draws a flat line across the whole window, and one with two or three readings draws a line joining them, which is too few to read a trend from.
Lec 1 · Zhou, B. · 109 of 120 enrolled
Freed 11 seats over about 1 month
How readings and alerts work
A reading and an hourly record are different things, and most charts here hold more of the second missing than present. dibs has taken readings of a section since long before it began keeping an hour-by-hour log of when it looked, so the empty stretches are dibs saying it was not keeping that log yet, not that it looked and found nothing. That is the age of the record and not a fault.
Seat counts show the last successful read, not live availability. Refreshes can be delayed by request limits or upstream interruptions. Watched sections receive priority, but an hourly update for every section is not guaranteed.
The registrar can change a seat count between our checks. Alerts depend on dibs detecting an opening and completing delivery; a quiet watch does not prove that no seat opened. Check the registrar before enrolling.