Cool

Good

The material is very useful and is presented well in the videos. The labs with only Jupyter Notebooks work well. The labs with CV Studio were not always successful. First, the instructions for using CV Studio in the assignments need to be updated and clarified. Once you get past the differences in the various sets of instructions and get something working with CV Studio, the labs with only the Jupyter notebooks work well. However, the labs in which you're asked to test your trained model with their "one-click app" do not. For these labs, you train the model with a Jupyter notebook (which worked well) and deploy the model in a one-click app, but when CV Studio says the one-click app is "ready", clicking the link lands you on a page saying that you'll be redirected to the webpage when it's ready but it NEVER redirects. I wish that I had known that from the start. I re-did these labs multiple times because I assumed that I was the one making mistakes. However, the final project appears to have been changed to use a Jupyter notebook instead of deploying a one-click app. If they re-work the labs either to include better instructions for using CV Studio while avoiding the one-click app or to avoid CV Studio all together, the course would be better.

The videos are high quality, however there was some kind of glitch that requires changing the video setting from 720p to something lower, then back, so that you can at least see the video and understand the content - Coursera should have this fixed. A lot of the labs need updating, to match the hosted support systems. There were a lot of deprecation warnings when running the notebook labs and the screen shots students are supposed to use for reference in the IBM labs are out of date with the updated software running there, leaving students to guess what to do next (unrelated to the assignment at hand). The applications would not publish as expected on the IBM Watson site either -- had to do workarounds and complete the final assignment by just running classifications straight out of the Jupyter notebook, and not through the web app as designed. Final assignment was also not clear on what images to process for the final grade and surprises students with much more additional work to search for images online that cannot be controlled or verified for the final grade -- which is 25%. This experience could have been done better. Please redesign it.

While the course is mostly good, it does have some pretty large issues that might cause problems: 1. A large portion of the material is on a different app, which unfortunately is not intuitive to use. It took too long to realise that the code snippets can be run natively in the text file they are presented in. 2. The videos go from 0 to 100 a bit too quickly. It feels like the teacher just wants to get the facts out as fast as possible and doesn't attempt to make sure everything is understood. 3. The grading and content don't match. The tests ask pretty simple questions about roughly 5% of the material and the remaining 95% are untouched. It feels like that 95% is surplus. All in all, the course is a pretty good intro into the basic theory, but not a good way to learn how to do the stuff yourself unless you the exact same tools as the material is in.

I took this course as part of IBM AI Engineering certification. The course should have been done after course for PyTorch as more than 50% of labs are with it. Initial labs are okish, latter ones are much better. But learning videos are with audio that is AI generated, theoretical explanations are thin and could better cover topics from the labs (some are not even covered at all). Luckily there is YouTube and internet. Course is more of exercise than really challenging. Peer graded assignment is interesting as focus is not writing code, but finding examples that do not work and how they can potentially be made to work.

Half of this course was amazing (modules 1,2 and 5) but the other half (modules 3,4 and 6) wasn't so good.

There is a huge gap between the didatic the good and the bad halves. Some of the LABs were amazing, but at the same time, there were LABs with procedures (or codes) that were not explained and even the team couldn't explain why some steps were taken.

The final task was very bad... the Open CV didn't work at all and a lot of people got "black screened" with no return.

In general, the course was good, but could be a lot better.

CV Studio is crap. It's difficult to use and it's integration with Jupyter is horrible. This course would be far better if CV Studio was not included. Even better if you stop using Jupyter completely. It would more realistic to just provide the source code I can run on my machine. I'll never use Jupyter in a real-world project. Also - Several of the coding exercises did not run in Jupyter as written; I had to make code changes just to get it working.

Interesting topics covered. But the presentation (explaination) can be lacking depth. Simple concepts were heavily empasized wile complex ones were flawn through. Honesly, there are videos on youtube for free that can explain the topics better. Thus, one needs to augment the learning process here on Coursera with outisde materials. The course is not comprehensive. Lab are good but not perfect. The machine learning course was much better.

The positive: access to a sophisticated platform hosted by IBM.

The negative: a general lack of attention by the course runners, including errors in notebooks that persisted for months on end.

Also, many of the exercises were hardly challenging or engaging. By this I mean a student could open a notebook, click on every cell, and get credit without having learned much at all.

Probably the less structured course in the specialization. Jupyter notebooks do not match with video content nor with the final assignment. Videos are very general and reused from some presentation. Compared to rest of the courses in this Specialization this one has surprisingly not much of actual content in it and also surprisingly not well-organized.

The subject is very interesting and theroy lessons are well done. However, the practical side is chaotic in sense that the labs were not tested: it is practically impossible to foollow up the steps and understand something as the screenshots and instructions do not reflect the real actual UI in IBM Watson Studio.

The lectures were good in introducing AI concepts and image processing. The lab software, CV Studio, DID NOT work and I was given credit because I completed the labs to as far as I could get. This was very unsatisifying. I would not recommend this course or any other IBM courses that use the CV Studio for labs.

Its a really good course but it needs to be updated because the Watson UI has changed a lot and have to fix some issues with the jupiter labs that makes it impossible to finish the final assignment. Besides that I really liked the possibility of sharing your webpage with your visual recognition app.

It gives a good overlook about computer vision and machine learning. I didn't like that the projects consist on simply executing already written scripts, which makes it difficult to learn do these things by yourself. If you want to become an indepenendent user, I wouldn't recommend it.

The course is super and very understandable and summarized to the essentials. But the compulsion to use the IBM services I find not good. Nothing works, I could not register for the cloud after days and thus not start with the final task.

There are too many incorrect errors in the video. Not only video but most of the time the lab was not working and when Lab started working the model could not be made available. Overall it was not a great experience.

Although learnt alot the course seemed somewhat disjointed with a couple of syllabus seeming missing (promotion to the web). the final assignment was a not as challenging as I would have expected.

The final practical was difficult to navigate as Watson threw up errors. I'm left in limbo and can't complete because there is no one else's submissions to review.

Pros: Excellent Content

Cons:

Information overload cut this in half or exten it to 8 weeks

Lack of slides, labs failed and were tied to IBM tools that also failed.

A good introduction to the basics of Computer Vision. However, the assignment platform has many technical issues and it is frustrating to submit the assignment

Please can you update the steps in the python code to deploy the web application. I think some steps haven't been updated such as the python version

The course is good but outdated. Many stuff need to be edited, it has been difficult for me to solve the problems and bugs of the course myself.

Course Says it is beginner friendly but it is some what of hard for beginners who don't have any knowledge in neural networks.

I think this course could use a little more balancing between the huge amount of information, and the easy practicals.