Great course for kickoff into the world of CNN's. Gives a nice overview of existing architectures and certain applications of CNN's as well as giving some solid background in how they work internally.
This is very intensive and wonderful course on CNN. No other course in the MOOC world can be compared to this course's capability of simplifying complex concepts and visualizing them to get intuition.
автор: Karol K•
Issue with triplet loss function shouldn't happen. I had to remove "axis = -1" in order to pass grader even though function had produced wrong answer!!!
автор: Дмитрий Х•
There are a lot of issues with programming assignments grader (I've spent one hour to complete assignment and two days to make a grader to get it)
автор: Roel H•
The programming assignments contain bugs. Also the jupyter notebook kept on shutting down thus slowing down the learning process quite a bit :-(
автор: Kalana A•
Certain Parts are not that much clear. Specially like in the triplet loss function, until the coding was done the real procedure was not clear.
автор: Kanishka D•
the assignment setup and graders are not updated after reporting issues several times which caused a great deal of frustration among students.
автор: Félix P G•
The last exercise it was a litle annoyng, it took me almost five days to figure out how to solve the face recognition because a grader fault.
автор: Sergio B•
I enjoyed the courses but I would like more practice, maybe a different module or examples aimed to help you define and optimize our models
автор: Serkan Ö•
There were repeats in the videos🤔 Also the answers to quizzes are not visible. If these would have existed, 5 stars would be reasonable.
автор: Rüveyda K•
Sometimes it was very difficult to understand lecturer because of his accent, but apart from that, assignments and lessons were helpful
автор: Stephen D•
This course is pretty good. Some things are not explained as well as Prof. Ng typically explains things, especially in the last week.
автор: Carol S•
The Neural Style Transfer notebook seems to have makes it difficult in the last panel to access the generated_image global variable.
The assignments need to be reworked as they are quite confusing and the grading system is flawed especially for the last assignment.
автор: Jnana R D•
More simple lectures with illustrations required and also graders need to fixed. Had a lot of time wasted because of buggy graders
автор: NEEL V•
It has very less explanation about working of back propagation of convolution network,
plus it can explain YOLO in much better way
автор: Aoun L•
The course is great but the assessments and grading is terrible, so many particularities and repetition that does not make sense.
автор: Sudhanshu D•
Week 3, exercise 2 is very buggy. Couldn't have completed it without the discussion forum. Kindly fix it for the future learners
автор: Ankit R•
Found it really difficult to submit programming assignments, at times the jupyter notebooks were not at all responsive.....
автор: Gowdhaman S•
Course content was good but lack with hands-on projects. It would be really helpful if the team could add capstone project.
автор: vishwanathan r•
There should be a way through which folks can download the entire zipped contents of the programming and also the lessons.
автор: Carmine M•
Very interesting and with high quality material. It could be improved by adding more tutorials about the frameworks used.
автор: Corbin C•
Poorly edited videos, poorly worded (bordering incorrect) quiz questions, buggy notebook. Good, useful content, though.
автор: Bence K•
The content of the course was perfect. But the support for the issues and the feedback for the forum threads is awful.
автор: Siarhei K•
For the YOLO part would be nice to have explanation for how to set up training set and train your own object detector.
автор: Junchen F•
The homework portion of CNN is underwhelming. We are asked to play around the core algorithm but not the core itself.
автор: Claudio T•
The content of the course is a lot of fun. I loved this module. But unfortunately the grade engine wasn't work right.