Positives:

1) Well designed course that takes you through the concepts of CNNs step by step and introduces cutting edge state-of-art applications based on it.

2) As always well prepared lectures effectively deliver the course material.

Negatives:

1) Course lectures should have covered overviews of actual models used in assignments (YOLO for object detection, Inception network for face recognition..) and the actual cost functions that were used to train them. That would have helped a lot in getting more practical real life feel helping user community a lot.

The content is great. The best so far in the DL specialization perhaps but I lost a lot of time in the last week's assignment where the grader was prompting wrong output in spite having written a program that gave the correct output. That was very frustrating and the worst part is I still don't understand why that was happening(I got full score after submitting the same program for the 10th time) and even jupyter notebook took ages to load(my net speed is 140MBPS). Apart from this, it was a really nice course and the experience was very satisfying.

Like the others, a fantastic course. Some of the videos and exercises seem a little underprepared, and require more time examining the discussion forums than the first three courses. For example the NST tutorial appears to require using np.square rather than tf.square to obtain expected results. This is not documented, and obtaining other results may result in passing, but it is unclear the ConvNet is working as it should. However the course covers current and quite advanced topics extremely clearly, and includes great links to original papers.

The content is superb, but the realisation of the course seems a bit rushed in comparison to the previous courses in the specialisation. The editing of the videos has many issues (fragments that were meant to be cut out are left in the lectures), there are many typos in the notebooks and the references for documentation are outdated. In one case the grader of the notebook has an unexpected mistake built in (it expects one rectangle area to have a negative value and gives a 0/10 grade when you try to prepare the code for such edge case).

Everything was absolutly great and undoutdetly quite usefull , I have one complaint though , if I had a problem in assignments( mostly related to grader output like U-net segmenation which I passed all the tests successfully but somehow my grade turned out 66) and I 've put my problem in discussion no mentor would answer it and if they did it was always the same "The topic has been moved somewhere that I don't know"

anywho it was great thanks for allowing me to take this course , it really helped me I hope I could return your favor

Very good course. I only wish there were even more examples with harder homework. I did every homework in less than an hour and felt like I still couldn't reproduce much on my own after taking a class. The lectures are great and cover most of the material you need. But as programming assignments go, it is still a very introductory class and you are definitely not ready to write much on your own afterwards. But I think they can improve a lot with a few more homework jupyter notebooks that are more advanced that their current ones.

Course provides good overview over state-of-the-art techniques in computer vision. The lectures are mostly clear although week 3 and 4 do lack some explanations on how these systems are trained. E.g. the style transfer lecture should emphasize that optimizer acts on the pixels of the generated input image using the without altering the weights of the network. In terms of the programming exercises, I would prefer to work through the code to better understand the structure, rather than just filling in bits. Still a great course!

Great explanation of advanced topics in deep learning and computer vision. This course deepened my understanding of convolutional neural networks in significant ways. The videos could use a bit more editing to remove repeated phrases, but it didn't distract too much from the learning. The projects are very good at giving hands on experience with the concepts and the tools. It would be great if they could be updated for newer versions of Kera and Tensorflow. Overall, I would recommend this course to anyone interested in CNNs.

Excellent lectures with really engaging explanations and examples. The programming exercises are also really well structured and require real work and understanding where code is required. It is very exciting when you get the models to run and produce a result. One concern: My sense is that we are moving a little fast on the overall process of solving a problem. That is, the programming exercises are so well structured that I am not confident that I could solve the problem without the exercise frameworks.

The course content in the video lectures is very good -- all the visual explanations and Andrew Ng explanation is easily understandable. But as far as assignments are concerned, many of the functions are readily implemented and just called for use. It would have been good if the student is guided in implementation from scratch. Though a top view understanding from theory to implementation is obtained with solving the assignments, it would have been more profound if everything could be implemented from scratch.

The material in the course is very good, specially the notebook exercises.

There are some technical aspects that prevent me from giving it 5 stars.

Namely, the english captions of the videos have too many mistakes, considering how easily they can be fixed.

I don't see the point of having an extra 'Reading'material with corrections for the videos. Why not adjust the video once and for all?

Finally, the number of clippings and strange cuts and jumps in the videos have increased with respect to previous courses.

Comprehensive course with a huge dose of knowledge about different CNN architectures, image recognition and neural style transfer as the last assignment. The lecturer teaches a lot about the theoretical part and programming assignments are demanding. However, after completing the course, I don't feel confident enough in using Tensorflow as there are some exercises that I've finished mostly by trial and error without actually understanding why they worked with so many other questions about theory in my mind.

Overall a very good course. Assignments have errors in the code. This is documented in the discussion centre going back a couple of years, mentors help explain, but mentors cannot edit to fix the code, and the course supervisors have long since disappeared. So you have to submit incorrect code to pass, then fix the code for your personal private code store - as the fixed code generates the correct numerical answers that unfortunately do not match the numerical answers that the grader requires to pass you!

Excellent survey of the area. However, programming exercises vary frustratingly between cut-and-paste and obscure tricks that require burrowing through the forum. Long-term value significantly reduced by the apparently intentional decision to not support efficient download of needed material. Unfortunately, Coursera's notebooks are also not stable, routinely resulting in lost work, followed by multiple tries to log back in. Love the content, but the rest of it seriously reduces productivity of study time.

The course material is very interesting, but also somewhat hard. It takes everything we learned in the previous 2 courses to the next level.

This is a good thing! However, the programming exercises do not really require the student to understand much. In most of the exercises I copy pasted from the examples and used some trial and error. Contrary to the previous courses, I feel that the exercises were something I did to pass the course - and they didn't really help me understand the material better.

Overall, this course clearly explained the basics of CNNs. However, the neural style transfer network could use more details. For example, after calculating the loss function which layers are updated in the network? Is it all of them or only the ones used to find the style image. Also, 'a_G' not being evaluated in the assignment wasn't clearly explained. Lastly, please have a couple of more assignments on tensorflow. I've done the earlier courses but they don't give enough of an understanding.

As usual, very informative and challenging at times. Andrew does a great job of introducing complex topics. I find that some of the quiz questions are ambiguous and result in the reader selecting the wrong answer. The hands-on assignments are generally of the right level, although a longer introduction to Tensor Flow would be useful. The delayed execution nature of TF can be confusing at times and it is not always clear whether the formula should be using TF functions or Numpy functions.

Significantly more difficult than the other courses in this specialization, but a good course none the less. Lecture quality is excellent, and the programming exercises are good. The reason why I'm giving it 4 stars instead of 5 is because they really need to spend at least 1 week in the course going over TensorFlow by itself. It is not an intuitive architecture and is almost like it's own language. I think students would get more out of this if they were better versed in TensorFlow.

I think the programming assignments were really good in hand holding and making students learn. As part of the course, I wish there is a open ended project to achieve an accuracy of X% on a given dataset. This could help challenge students to have some comprehensive(whole model building) experience. I understand that this increases the course completion time as a trade off. But I think an open ended project with a predefined dataset would have added well rounded value to students.

Awesome course - a lot of material and complexity covered with good examples and assignments to understand the range of architectures of CNNs and their applications. Really enjoyed learning.. But working on programming assignments was frustrating given the grader shenanigans.. Coursera service on this is not great and sometimes tips from Mentors didn't really help. Greater attention must be paid to ensure that learners have a consistent experience with programming assignments.

Another great course in the series. The later labs were difficult, some additional time in the videos on TensorFlow concepts would be helpful, hit some frustrating points in the weeks 3 and 4 labs. Also helps to have background with linear algebra (or it's a tough intro and notice to study up on the stuff!) Overall another awesome survey of the state of the art, lots of practical advice along the way, the links and discussions to the underlying papers were great.

One of the most clear convolution explanations ever! Loved the mostly recent algos discussed, too bad all important papers don't have breakdowns like that!

Negative:

-Video editing has gotten worse in later courses, lots of areas where Andrew clearly thought he would be edited so he repeated himself

-Quiz feedback is non-existent. There should be blurbs explaining why answers are right and wrong, instead of just saying it is wrong or right with no supplemental text

Nice lectures and exercises.

Unfortunately, although there are some problems with some expected results:

1/ Face Recognition: The Grading Process has some bugs, issues if TipleLoss function

2/ StyleTransfer: model_nn fonction give wrong "expected value" by using indication to define cost. It seems like the problem come from tf.square that not gives the same results as np.square (correct value)

Please fix these issues, then then lecture will be perfect.

Thank you

Prof. Ng is a very good teacher and the course is content-rich and well organized, but there are two things that could be improved. First, there are many hesitations and reformulations that could be removed from the videos. Second, and most importantly, the assignments are a bit too easy. The answers are almost given in the questions, and if you know how to translate equations in Python you can manage to pass without really understanding what you're doing.

Excellent course, but some of the most technically difficult material I've ever encountered. Gave me a solid understanding of the theory behind CNNs and their applications. However, it doesn't go deep into how to program using tensorflow and keras and holds your hand through most of the assignments. Overall I think you'd have to supplement this course with some keras tutorials and lots of practice to be able to implement any of the assignments in the wild