Avis et commentaires pour d'étudiants pour Robotics: Perception par Université de Pennsylvanie

The english of the teachers is difficult to understand. Is good to know that you can become a professor of Penn university with that english, but to be in front of a camere they should select somebody understandable.

The level of the course is way too difficult. I have a bachelor in computer science, a master in computer science and a bachelor in mathematics, so I don't think the problem is the material, but how it's teached. You are supposed to solve very complicated assigments just with the videos and that's a too big challenge. They don't spend the time with examples and detailed approach.

Thanks god that the course has been active for several years and there's enough help in the forums. By the way, people in the forums complaint about the poor support of the teachers and assistants in the forums.

Very bad course in general.

Sloppy, unorganized, and poorly presented. Try the similar class on EdX and YouTube for a much better and clear explanation and presentation.

The least comprehensive course I ever took. It needs very good prerequisite about computer vision. And the demostration was very poor.

The assignments and weekly work is by far much more than the hours estimated. Finally I feel as I didn't learn anything.

This is quite challenging course. So far, this is the course with the largest amount of material, I wish the class will be split into two courses.

Great course for those who wants to understand how classical SLAM systems work. I think it would be a bit more practical if the assignments were made in python.

For Computer Vision enthusiast who wants to learn about Multiple View Geometry, this is the best beginner course

The content is very good but the lecture presentation and structure could use improvement. Assignment instructions are not as clear as they could be.

An *excellent* but tough course! I really struggled with it because it's a little out of my expertise (I'm an engineer with two master's degrees and equivalent of almost a half degree in math, , but descriptions of the perspective manipulations are novel to me). I managed to do the work by watching the videos several times and reading the forums. I prolonged the course by continuing to pay the fee for quite a few months. This last assignment was very difficult, but I finally sat down and worked through it - scored 50%, then maybe 67%, and then 84% (passed) . I'm disappointed that the course seemed to accept that as the final submission, and I think I'm being forcibly "graduated" and advanced to the next class. :) I actually want to get the final NonLinearTriangulation part of the routine fully working and maybe earn 100%, but I guess I will do that on my own time. Also, I'd like to comment on the forums as to some specific methods I used to successfully complete the last assignment. I am looking forward to the next class (5 of 6 ) being much easier for me, as the material looks a lot like things I've already covered.

Excellent organization and presentation of the course material, and very prompt responses from the teaching staff on the message boards.

Very good course. The only thing that I would suggest is an higher precence of moderators in the forum. It would be very appreciated.

good materials for visual SLAM

A BRILLIANT, BRILLIANT COURSE, the teachers put A LOT of effort into making the lecture slides and videos. Everything was explained multiple times such that the student understands it better. Also, computer vision, especially geometric vision is difficult to understand without a proper background in linear algebra, but the teachers' explanation was enough to fill the gaps so that even someone with only a minimalistic knowledge of linear algebra was able to consume the content.

The exercises were a class apart, they were very well structured with tangible results at the end of each, And each of the exercises brought together the key points of the lectures, so that the student could easily implement them in code and test out the algorithm.

Last but the not the least, the community was very active with the teaching assistants pitching in wherever necessary, in particular, Stephen did a great job of understanding the issues students were facing and taking appropriate action.

All in All, a very well structured course to jump start one's career into computer vision.

It is certainly the most comprehensive course in computer vision that can be provided in a span of four weeks. It is much time consuming compared to the other four courses (I have done this at the end); however, each and every bit of it is worth it. The teaching is incredible, especially, Prof. Shi's teaching includes intuition and physical interpretation which helps in appreciating the equations much more. The assignments equally match with the lecture content. Trust me, by the end of four weeks you will be comfortable in reading and understanding papers in visual SLAM, pose estimation, etc.

A small suggestion: in a few lectures, for instance in SIFT lecture, Prof. Daniilidis is not shown in the screen whereas his actions are necessary to better understand the content in the slides.

The Professors and the TAs have done a commendable job and thank you all for this course.

This course is excellent: lots of things covered in depth, learning curve is high, detailed explanations with lots of examples; If you want to ramp up quickly on Structure for Motion or Visual Odometry, Visual SLAM, this is highly recommended. But be prepared to put some real effort in this demanding course. Overall one of the best MOOC I took. Programming assignments and especially the last one are *very* interesting. It's great to have such courses that are available for everybody. Pre-requisites are linear algebra (eigenvalues, eigenvectors, Jacobian, Hessian ...) and familiarity with matlab (but people familiar with numpy should easily ramp up). For people not familiar with matlab there are also some very nice matlab tutorials in the resources. Highly recommended.

Great Deal of Math.

Prof. Shi's lectures on math guides me through this course. Whenever he shows up in the video, I know he will give me almost everything I need to solve the problems.

Really Intensive and rewarding.

The programming assignment is not that difficult if we have understood the meaning of the equations on the slide.

But the math is not easy. Though Prof. Shi has been giving the lectures in a rather reasonable pace, I still have to pause the videos for quite a long time to follow him on math.

I WILL NEVER FORGET SVD AFTER THIS COURSE. AMAZING!

Hope Coursera can offer more intensive courses like this. Really like courses going in the order of advanced math - algorithm - practice.

This is the -hands down- best course within the Robotics Specialization. It is educating as well as entertaining (well, as far as a mooc about robotic perception can be, but I loved it!) and you will learn A LOT, if you don't give up and try hard. Because this course is not easy at all. Its not for beginner and sometimes I had the impression that its neither for people with a somewhat intermediate level of engineering and/or mathematical understanding. I struggled about 40 hours with the final project, but in the end I managed to finish successfully! Thanks to the staff for this very cool learning experience!

As a standalone course - incredible. Lots of content, in detailed guides through maths. You seem to have been structural, but I still go lost every time. Overall impression is very chaotic. Maybe more summaries could help. Also separating videos into 'guide through maths' and ' guide through reasoning/theory/motivation' could help

As a part of specialization: This course has more content than the previous 3 courses combined. This took me by surprise and it actually took me 3 months to finish this course. Maybe you could saparate some stuff to optional and required?

1-Fundamental Course in Computer Vision. 2-Very helpful and comprehensive course material. 3-Some repetition by instructors. 4-Very interactive slides and Good use of colors for expaling linear algebraic derivations. 5-Comprehensive programming assignments.

Things to improve upon: 1-Sometimes its difficult to comprehend verbal delivery because instructors are not native English speakers.

Remarks: I loved this course and will recommend for beginners in Computer Vision. But keep in mind that it is very demanding course.

This course has everything it takes to understand SLAM. The instructors have worked very hard that they start with the basic concepts of perception and work all the way up to components of current SLAM. You will be very well able to appreciate the current code base on ORB-SLAM as code does not involves derivation and this course covers all the linear algebra behind it. Hats off to professors for not skimming through concepts and making sure they convince you without needing to refer to external resources.

Amazing course on computer vision geometry. I did not expect much from this course to begin with, I was expecting a walk along the park instead of deep knowledge I wanted. But! it exceeded my expectations from week 1 and forced me to go look for other materials and refresh my rusty linear algebra to pass the challenging quizzes and assignments. When Professor Shi talks conceptually, it is very easy to picture. The maths just flow naturally once the concepts are understood.

I Highly recommend!

This course is definitely worth learning if you are interested in computer vision or robotics perceptions! There are some minor flaws in the lectures slides, but it doesn't seriously effect the learning experience. I would recommend this course to people who have some basic knowledge about computer vision (e.g. camera calibration, coordinate transformation, affine/rigid transforms, linear solution of structure from motion). Otherwise, the latter part of this course could be a bit difficult.

I cannot believe I finished it! This course involves a lot of mathematics and requires a lot of brain works. Things taught in this course are what I need to know in my work as a self-driving vehicle software engineer. Both instructors are amazing and know what they are teaching. Especially, Professor Jianbo Shi explain the math equations clearly in great details. I would recommend this course to whoever wants to work on computer vision/ visual perception module of self-driving cars/robots.

This course is interesting and very thorough. Some concepts of robot perception are explained in detail, with a focus on perception based on 2D vision. The videos are clear and there is a great number of quizzes and Matlab programming to improve your practical understanding of the topic. Be warned, though, that this course takes longer than 4 weeks in fact due to the numerous and long lectures.

This is a Coursera course with the richest contents I ever had. Very glad to have learned so much in robotic perceptions. Thanks so much to Prof Daniilidis and Prof Shi. it is challenging but also very useful and helpful for further study or research. TAs are also very good helping lots of students. Love this class. Thank you all!