Excellent course once again. There were some nuances in the programming exercises that included some time consuming overhead for troubleshooting (thus 4 stars). Nevertheless, would highly recommend to anyone aiming to dive deeper into convolutional neural nets.

Thanks again, another useful and well thought-through course. Some tidying of the videos and checking of the scripts instructions (especially non-mandatory steps) would polish this course. Overall it flows well and contains very useful insight and intuition.

Great course for beginners but disengagement of the tutors for some concept related questions and very small number of typos on the slides are little drawbacks. But definitely encourage everyone to take this course for getting the building blocks of CNN right.

Concept of anchor boxes was not clear. Hopefully, you can add some more clarification for future students. How are anchor box dimensions determined? What happens to objects that are near the camera vs those that are far away? Do anchor boxes scale accordingly?

Learned a lot, obviously, but felt like I had to look for answers more actively than previous courses. Obviously, the positive side is that you get trained at debugging as well and at searching the internet. Being self-taught anyway this somehow felt familiar.

grading for assignments seems buggy, and mismatched to hints on the discussion forums.

Slight differences in an arg to a function can result in a correct result in notebook but failing assignment. This was un-intuitive and confusing. Content was good otherwise

Would give it 5 stars were it not for a) the grader problem in the face recognition exercise and b) some of the obscure tensorflow in the NST exercise.

But all in all prof Ng is brilliant and the way the course is set up is very intelligent and challenging.

Subject matter is interesting and Andrew does a great job (as usual) to teach it. I do wish there would have been more practical examples in the earlier topic in order to make it easier to get a intuition for what the different network architectures do.

It is not as detailed as previous courses, but it is a good course. I wish it would have more details abouts how to see what is doing a convNet and how to see inter-layers outputs more detailed, and also how to tune the network with conv layers. Thanks!

Andrew makes the concepts very understandable. The way that he put assignments together helps. I wish there was more references to explain some ideas in more detail. This course has room for being expanded a little bit. Thank you for all the good work.

Generally it is great to learn concepts related to image classification/object detection and etc. Some details of certain areas are missing, such as how to now the bx, by, bw, bh in YOLO algorithm. Have to move on to learn details from other channels.

The video lectures are very good at outlining the concepts. The programming assignments with the jupyter notebooks are nicely done, but they do not go into much depth. Hence, the course mainly gives you theoretical knowledge than practical experience.

This one proved to be the most challenging for me in the specialization, especially the programming assignments. While the concepts are very very well explained by Andrew, the application (though well guided) required more trial and error on my end.

I really liked the course but there are many tiny errors in some of the videos, which they have fixed in a following article but I got stuck in a couple videos because of those errors and later saw the article. Otherwise the course is really nice.

Presentation of material is fantastic, but there were A LOT of technical problems with the grader that led to a lot of wasted time and frustration. Very good course, but please work to update the grader issues so future offerings are less buggy!

This course, out of all of them, seemed to have the most grader issues. Several times I had functional code with the right answers, but it got marked wrong and I had to hunt through the discussion boards to find out how to do it the "right way".

Excellent Course content and very aptly put tutorials .Course completetion is just a milestone and You keep going again and again these materials (even though you have gained certificate )you will learn something new. Thanks for this series !!

Great course, but some discrepancy between face recognition/verification notebook and the grader make this impossible to get full grade (I had to check in the forum and enter an answer giving a result not corresponding to the expected output).

I really wish they didn't provide so much boiler-plate code. It seems to detract from understanding the programming assignments fully. Yes, building the whole thing yourself would take 2-3 times more effort, but the end result would be better.

Tensorflow tutorial is too short. Although i finished this course, i still dont know how to use tensorflow to build my own nn and how tensorflow actually works. This course need update the tensorflow to v2, and provide more in-depth content.

Very nice course about a special type of Neural Network.

The course materials are really good as well as the teaching style of Andrew.

The only minor point is that the programming exercises are too easy with only "fil in the gap" challenge.

The course is good. Well explained.

The videos need some editing, sometimes speech is repeated which doesn't help with concentration.

Also, the forums need fixing, for someone used to Stack Overflow (and others) the forums work really bad.

Need more quick help from discussion forum since it is very important to understand the usage and working of components in the code. The course is superb in theoretical part but I felt I needed more assistance in programming exercises.

The contents are good, however the exercises includes too many errors and it takes too much time to read all the discussions to find out what the hack is. You should make sure the exercises are working smoothly before you publish them.

Since I have a computer vision background I was expecting much more challenges at this points when doing the pratical assignments. The explanations and intuititions about ConvNets are awesome , and this why I am giving 4 instead of 3.