This course really gave feel of deep learning. The way Andrew Ng taught and content of course is preciously valuable. star less because the language Keras and Tensor Flow where I felt difficulty. You should provide language study material with examples. I hope you would implement it. It will help to work with language with ease.

The content of the lectures is great, introducing state-of-the-art solutions in a way everyone can understand. However the grader for the assignments needs fixing. Many times I had a correct solution, spent a lot of time trying to fix it only to discover via the discussion forum that an incorrect solution passes the grader ;/

A great introduction to convolutional networks. Highly recommended to learn about the cutting edge scale and depth. Minor suggestion: The Jupiter interface seems to be buggy on chrome (mac). Doing a save and check point can help relieve this problem (at least re-coding wont be needed). It would be nice if a fix is available.

Excellent videos and quizzes to learn or review CNN concepts.

The notebooks however should be refresh using modern frameworks (Tensorflow2 or Pytorch). The explanations inside are great, but there are also too guided: more freedom should be given to implement methods with only rigid formats regarding submission evaluation.

Excellent as usual lectures from Anrdew Ng. But at some videos some key points are missed like Neuro Style transfer is transfer learning and we use pre-trained CNN. It would be good to have some data flow drawing. YOu get it later from exercise but not from video at once. Exersises are less accurate than previous courses.

It was my first specialisation course , i felt great and learn lot from it , i also get a good hang of CNN and CV from basic to core , the only downside was the assessment system it was too confusing at first and i still have not full confidence in building great convolutional models but its a great start for any amateur

The course does cover fair amounts of basics very well. But the course content just provides a starting point and I feel it lacks depth. The course assignments are nice and give a good platform to implement what was learned in that week but it was more of filling in the blanks rather than building a full-fledged system.

Feel this course's assignment is not quite heavy on why people build the CNN that way. But it tells how the thing is. I would like a "Future work" section to give some inception on the future of CNN and limitation in terms of image orientation, content, what the deep learning cannot solve right now for image based work.

A few small grader errors, and some implementation details were glossed over and not fully explained (had to look into the util code to fully understand what was going on), but otherwise a very good course, just not quite as polished as the first 3 (perhaps understandable given the increased complexity of the material).

This was a great course. I really enjoyed the lectures. Andrew NG is a great teacher for sure. The only reason why I give the course 4star is I felt that while doing assignments somebody holding my hand and helping me too much. I mean practicing part is a bit weak in my opinion. Other than that these courses are great.

Frustrated since in week 4 face recognition notebook, you need to put the "wrong" code there which did not match the given output to "cheat" the grader to get points. But overall, this one is the best in the first four courses in the Deep Learning Specialization. Looking forward to see the fifth course sequence models!

The only criticism I have is for week 3... I know YOLO algorithm is difficult to implement in a teaching environment, but still, I struggled a little bit to link all the pieces I need to get to object detection. I'll have to get back at the course contents on this one, but I believe it wasn't as clear as it could be.

The course was very good and offered a nice sample of the state-of-the-art CNN models. I would improve the explanations regarding TensorFlow. Understandably, exercises cannot ask you to train a model from scratch as it takes too long. But when reading already implemented TF code, it's hard to understand some steps.

The course offers some great insights into conv nets in a very easy to understand and concise format. The one thing that bothers me is that the programming assignments have a lot of stuff already coded and I get the impression that after doing them I don't manage to learn everything that was used in the notebooks.

The video/quizzes and practical assignments are outstanding.

On this run of this course, the practical assignment grader had too many issues, which meant wasting time trying to "fix" code that should have been graded as passing.

The forums proved invaluable to see how other student got around the grader issues.

It was a very insightful course as it let me understand some of the most recent papers in computer vision. I would suggest to improve the programming exercises as I observed that the exercises were a little too simple and some graders had a few bugs resulting in them recognizing wrong answers as correct ones.

As audio guy I wish to see some example on that domain as well beside image processing since Conv2D network are applied to process audio spectrograms. In any case Andrew Ng does (as always) an excellent job to present Machine Learning concepts giving intuitive clues without giving up to rigorous formulation.

Great theoretical content of CNNs. However it would be better if there was a section on the practical implementation of these models. Eg: how to connect to GPUs to train a model, more on loading and saving models from some open source website, using a loaded model how to train it to recognize something else,

Other than the Face Recognition Assignment (grader problems) the Course was a very good experience. Also I would have really liked if some more instructions in video form were provided for programming frameworks. But still the information was almost adequate. Looking forward to try my own models now. Cheers!

Very good, Overall I learned a lot about convolutional network and the meaning of computer vision. The downsides of this curse, was the first program assignment, and there were a lot o videos with errors, that got me back seeing all over again to understand what i was doing wrong in the program assignment.

The course content (and Andrew's teaching, of course) was very good, but I think there were times that the learner was suddenly dropped into deep TensorFlow / Keras waters without enough preparation / hints. If I hadn't had some prior TF / Keras experience, I don't know if I could have completed the course.

I didn't give 5 star as usual because I felt that the course materials and homework are less perfected than before. A few times I got stuck with homework and reviewing course videos again or reading the homework guidelines didn't help, so I need to go to the discuss forums to find out some of the nuances.

Good curse and Andrew is the best teacher for learning this topic. But there is not enough hands-on exercises. After 3 courses at this specialization I still don't understand well enough how to implement convolutional neural networks in TensorFlow or Keras. Perhaps it's better for very beginner students

As I have gone along in the deep learning specialization courses I am noticing more and more places where the video was not edited cleanly. For example, Andrew will repeat a sentence multiple times assuming his error will be edited out later, however there are many examples of these edits not happening

Fantastic, but it would be more reasonable if the courses from this specialization was more "hands-on". I feel like I can implement nothing by myself after this course without looking and stealing the code simultaneously. But other than that, I would love to recommend this course to any of my friends.