Good course but lot of technical issues

Course Material needs updated.

bad codes

A gargantuan amount of contents crammed into one small course: OpenCV, Numpy, Pillow, Pytorch, Scikit-learn, Normal Neural Networks, Convolutional Neural Networks, Hyper-parameters search, Fine-tuning, Deployment strategies. I must have forgotten something for sure. Okay for revision, if you have already learned the stuff elsewhere. Absolutely no chance you will remember any of it 1 week after completion otherwise.

A ton of projects, but all just click-through or copy-and-paste a-la Udemy style. One could absolutely understand none of the given code and easily end up with 100% grade from quizzes that only test basic knowledge or could be brute-forced. That is, if you are lucky and don't get docked marks during inadequate peer-assessment. All it takes is just one grumpy student to drop your grade into oblivion... without any feedback or reasons given... possibly because you previously flagged their totally empty submission.

And don't even get me started about IBM constantly pitching their Watson and CV Studio, which no one uses... Which doesn't even work half of the time. Good luck deploying your model on the last project!

Unfortunately, this was the third course in the certificate. If only it were the first one, I would have abandoned this cursed specialization like some dirty diapers.

I have participated in many courses on IBM but this is the first course that I feel very bad. The only good point is the first 2 weeks where they teach you the insights of Pillow and OpenCV. The rest is terrible. I feel like the knowledge is given without a clear explanation. The LAB Section is probably only for Python experts, you won't understand what they do at all if you haven't learned Pytorch before. Just run cell, look at the result, then run cell, look at the result, run cell, and so on... The final project is just like a slideshow no more no less, you follow the steps and show out results, done. You will not learn anything from data-processing step to refine-model step. I gave 2 Stars simply because I was inherently sympathetic to IBM before, but with this course itself, all was disappointment.

The lectures are very well made. They explain things clearly and contain good information. The projects need work. They need to be made to demand more interactivity from the students. The final project is absolutely horrible. Terrible design. Shame on who made it. I have peer reviewed 7 peoples projects and it is clear they have no idea what to do and I don't blame them. It needs to be completely redone. My suggestion is to do it like a kaggle competition. Give the students a dataset with images and labels and challenge them to develop a CV model to classify the images. Then give them an unlabeled dataset of images and ask them to submit the predictions for all of the unlabeled images in the dataset. Then find the error of their submissions. This could even be automated.

Content is very relevant and complete. But often hard to follow. Videos are narrated by a computer generated voice, making it hard to follow, and not well segmented into the different ideas and concepts presented. The labs are basically examples of code for the different models presented in the videos. There is only a minimal amount of own coding required. The graded exams are only multiple choice and no coding is required to pass. Everything I say i based on the first three weeks, then I dropped out. I was quite disappointed by this course.

This course has some of most interesting labs from all this certificate. However, they are full of bugs and outdated information. I spent hours searching for tips and workarounds in the foruns. By the way all my questions are still there waiting for answers.

Overall, it was a frustating experience mixed with some some interesting technologies and approaches.

The trainer sounds so disinterested in lectures. Lectures are monotonous. Also there are bugs in the CV Studio that makes it much harder to submit assignments on time. Course topics are otherwise good, but lots of scope to become better. I don't recommend buying this one.

Course content is good. Too many issues & errors in getting CV Studio working. Submitting assignment had been the most time consuming and frustrating task. People are struggling to submit the assignment and discussion forums are full with these system errors.

I have a lot of problems and issues with this course. And the last 2 questions of the final assignment are not very easy to understand what is it you want. My experience has been so bad I am just glad it is over

Good course, but terribly maintained Lab and Capstone project. Lab has been broken for two weeks. Coursera has however agreed to cancel the last monthly charge, but the labs need some serious TLC…

The cource content was a good overview of nerual networks and thier use in computer vision. The labs however where sloppily put together, and in most cases, not formatted correctly.

Some of the practice notebooks don't seem finished. CVStudio is very unreliable and doing the final project didn't really work (lots of comments in the forum but no action)

Computer vision material is very weak. It did not cover enough theoretical material or programming tools. Programming labs used pytourch wich is covered in the next course.

Little explanation and too much lab work. Doesn't get your curious about learning and out of date course. Some parts don't work correctly.

Great tech, great teachers, completely let down by course material that is almost always broken, out-of-date, or buggy.

Completey broken tools and incomplete directions make a great topic hard to follow

it was based on the tool which basically does not work

many problems with CVstudio and skills network lab

I was expecting the course to be focused on machine vision and image processing, however it only has about 1 hour on extremely basic open cv techniques that i could have learnt off youtube. the rest is machine learning and ai (which would be fine if done properly). The ai bits in it are brushed over in the videos, merely mentioned or spoken about with no context, then the labs are completely useless, entire portions of the setting up the models are done with no code annotation (apart from ' Download the files needed to build the model: ', several of the labs dont even work and have errors in them, the online jupyter note books crash every time you load one. the tests are a complete joke, completely pointless questions ('which of these is not classed as machine learning') and one of the tests had the same question twice (out of 5 questions).

IBM and whoever working for them who wrote this should be ashamed. its total crap.

Id be very angry if id paid for it myself.

If you actually want to learn something about machine learning go on youtube, DO NOT BOTHER WITH THIS

Worst course so far I've done in Coursera (and I think this one must be my tenth or eleventh). If you are interested in getting any kind of concepts or explanation on why and how computing vision works this course will be completely useless for you.

This is just a collection of pretty superficially explained recipes and I guess its only value is to be able to show off a pointless badge from the not so prestigious IBM (at least in my country Span, and I know what I'm talking about I've been an IT professional for more than tewnty years)

You will not learn anyhing from here but some familiarity with Jupyter notebooks, python and a lot of suffering with that crappy IBM online tool

One of the worst courses I've taken on Coursera (and I've taken a lot). Feels like it was abandoned years ago, and needs a serious tune-up for 2023.

Course is very high-level. Shows _what's_ happening, but very light on _why_.

The worst part are the labs:

Broken links to images.

Deprecated functions

IBM Lab environment is often broken

Some labs just can't be done because they're too broken.

Too much advertising for IBM's cloud, and it's a negative ad because the experience is awful.

And what is with the ENDLESS Captchas to get into a lab??? How many motorcycles does it take??? I having to click on 30+ images is excessive!

Very bad course, if you have a decent knowledge on the subject, then its not for you. On ther other hand if you are novice then you shouldnt start with that. However its greatest problem is that you must use a specific platform (CV studio) to train your models, which doesnt work properly, therefore you have a hard time to finish the course. Although i must congratulate coursera support center who helped me a lot to overcome this issue and claim my badge at the end.

I have no words for this course, but I'll try none the less to muster some with great pain:

1) Content is extremely superficial and important concepts are glossed over. You can argue that this is an "Introduction", but in my opinion it fails to introduce the topic adequately.

2) The course is extremely linked to an IBM platform. Let's be clear, it does not work.

So you are left with a useless course that you cannot complete. And no money back.