Jia-bin Huang

This professor has taught: CMSC426, CMSC733, CMSC800, CMSC848F, CMSC848K
Information Review
Jia-bin Huang
CMSC426

Anonymous
05/24/2024
This guy sucks, its been a week and a half since class is ended and we still have no grades in elms, how is this even allowed
Jia-bin Huang
CMSC426

Expecting an A
Anonymous
05/19/2024
All his lectures were watered down versions of other lectures from the internet. His ability to present new information is mid at best. He often felt disheveled and unprepared. His TAs also rarely provide adequate support. If you already like computer vision, the class will be fine. If it's a new topic that you are hoping to be guided into, this class might not be for you.
Jia-bin Huang
CMSC426

Anonymous
05/17/2024
All of the reviews so far (both positive and negative) are fairly accurate. Dr. Huang is extremely knowledgeable and proficient in Computer Vision, and seemed quite approachable and friendly. He would regularly connect the material with real-world, modern applications which made the class seem more practical. However, this class is EXTREMELY math-heavy. If you are a person who enjoys math and is interested in Computer Vision, I'm positive you will do well in the course. The highest math class I took at UMD was MATH240...2 years ago. I suffered heavily from my lack of practice and couldn't keep up during lectures. Since the material is so practical and hands-on, however, all of the projects were doable (with a lot of googling). We were allowed a cheat sheet on the midterm and plenty of extra credit opportunities were given. We also got seven days' worth of late passes. TAs were responsive on piazza and would do their best to help you. TLDR: interesting class, super math-heavy, nice professor, below average lecturing, thoroughly assess your own abilities before taking!
Jia-bin Huang
CMSC426

Expecting an A
Anonymous
05/11/2024
DONT TAKE THIS CLASS!!! you may be thinking "but you expect to get an A", yes but I can wholeheartedly tell you I dont know anything more about computer vision then i did when i entered. We learn ZERO code in class and this is NOT an exaggeration, the topic is simply too dense and to be learned in 1 semester. You sit in lecture to him talking to you as if you already know everything about the subject as he reads off slides that contains nothing but words and pictures of how images change from the algorithm used, while he "explains" the steps of the algorithm thats all he does. You spend at least 4 hrs just googling what functions must be used for the various steps, another 4 hrs trying to understanding how to use and connect the return values and then the last 58(a bit of an exaggeration)hrs trying to connect everything and debug. There were frequent and endless questions on piazza as people struggled with these projects to the point the professor stopped responding to posts and put it all on his TAs. The TAs were the only redeeming part of this class, they were awesome. Ive heard that SOME positive feedback from other professors, but if you care about your mental sanity do not take this guy.
Jia-bin Huang
CMSC426

Anonymous
05/02/2024
yk how some smart people should just not be teachers? well this is one of those. smart guy but horrendous teacher. lectures are entirely useless for projects and homework and his class is very disorganized. don't take.
Jia-bin Huang
CMSC426

Expecting a C+
soymilk
04/27/2024
TLDR: DO YOUR RESEARCH ABOUT COMPUTER VISION BEFORE TAKING THIS, LOOK AT PPTX FROM OTHER UNIVERSITIES THAT HAVE THIS COURSE. TAs AND GOOGLE WILL BE YOUR BEST FRIENDS I'm working on the last project right now (not the final project) and looking back at it, this class is very manageable. I barely got a 50 on the midterm, but I was able to complete all of the previous projects successfully. If you START EARLY and GO TO OFFICE HOURS, you will also be able to complete the projects on time (I started most of them a few days before the due date and submitted them late but I cranked each of them for around 2-4 days straight with help of TAs and finished them). I would only suggest taking this class if you are genuinely interested in the computer vision world. Look up textbooks and documents from previous years and from other schools. Jia-Bin goes into detail about concepts and lectures are mandatory, so if you don't want to fall asleep, do NOT take this course.
Jia-bin Huang
CMSC426

Expecting an A
Anonymous
04/25/2024
Very knowledgeable professor. Hard class with a lot of math but worth it.
Jia-bin Huang
CMSC426

Anonymous
04/23/2024
The lectures are really information-heavy with complex concepts BUT he never explains the practical application. The lectures are not helpful for the projects whatsoever. The projects are poorly designed with little to no guidance. Expect to spend significant time figuring out project goals and approaches on your own. He also makes course changes and does not update the syllabus promptly. Important announcements are made on Piazza, but not necessarily mentioned in class. Save yourself the headache and don't take this class with him.
Jia-bin Huang
CMSC426

Expecting an A
Anonymous
04/18/2024
I have never written a review before but I felt like I need to write one because the other reviews aren't giving the full picture. In college this is honestly my favorite class I've ever taken and favorite professor (second only to Justin). Huang's explanations are always very articulate and clear. This course also has the most relevant material I have ever learned in a class. Some of the coolest projects are stitching panoramas with almost the same algorithm the iPhone uses and tracking key points in an image and using them to reconstruct a 3D point cloud and the camera motion. Huang is extremely smart and gladly answers any questions that come up during class. I do warn you, this is a very math heavy course. I have a strong background in math and came in with MVC and two lingalg courses. I admit in the second half of the semester some of the math went over my head. That being said, it didn't really take away from the class at all as the theory behind the algorithms still made sense. Anyone who is good at math will definitely enjoy this course and do fine. I would recommend taking MATH 401 though to get more out of this course. If this is your first 400 level CS class (it was for me), it might be a little rough at the start but if this isn't then this course shouldn't be a problem. Also adding on to the point that understanding 100% of the math isn't crucial is that the midterm was fully conceptual, no code, no math. Another thing that he does well is he spends the first 10 minutes of every lecture reviewing the material from last class in a slightly different format with slightly different slides and this always helps everything click. Every now and then he'll do an even bigger review of the last couple classes or even from the start of the semester to show how everything ties together. TLDR; Hard class but very fun and well taught, you'll do well if you're good at math
Jia-bin Huang
CMSC426

soymilk
03/28/2024
just dont even take this course. professor is cracked at comp vision but the lectures/slides just does not help with projects. students often just end up lost and have to ask piazza how to even begin. google has been my best friend so far...
Jia-bin Huang
CMSC426

Anonymous
03/14/2024
He seems really smart but also doesn't give a crap about the class and just yaps and yaps and yaps without explaining anything. Basically the lectures are very high level concepts and he doesn't explain how to actually do any of the things he talks about beyond putting a huge, abstract algorithm/summation up on the board (without explaining) for 5 seconds. It's kind of the common L of professors who are really smart but don't know how to teach and only care about their research. The problem is that his lectures are so much of an information overload that its not really worth trying to figure out what he's talking about, because even if you do, the high level concepts aren't all that useful for projects. Speaking of projects, some of the worst designed I've seen in my time here. Little to no directions ever, and despite the TAs being smart and helpful, they usually can't debug with you because everybody does things in exceptionally different ways and it is impossible to unit test when everything is tracking hundreds of points across tons of iterations across tons of images. Most things in this class rely on packages, but often times some things just don't work in the way you need them to and its not really possible (feasible) to figure out why in a 2 week project (ie don't use numpy.gradient for gradients, it gives garbage) . After the first project dropped lecture attendance went down like 80% because everybody realized how useless they were with respect to what the projects ask of us. He then made class required (which is fine I don't mind) and added a midterm mid-semester (fine) but he never announces these things beyond piazza posts and doesn't update the syllabus. He makes large scale course changes and then weeks later talks about these things like they've been known (this isn't a huge issue, cause yeah, we can check piazza and whatnot, but still just a poorly managed course if he has to update the grade breakdown many times mid-semester).
Jia-bin Huang
CMSC426

Anonymous
02/28/2024
Lectures are huge information dumps that can get overwhelming. We don't learn the "why" behind most topics and there is a huge disconnect between lecture information and implementing the code on homeworks. There is zero guidance on how to start the homework and rules and guidelines will randomly pop up on piazza but only after someone asks a question about it. Attendance was made mandatory almost halfway through the semester because no one was showing up, but the reason people weren't showing up was that no one found lecture useful. Just a poorly run course in general.