This last course is much more challenging than the prior four, but provides a very good launch pad for taking what you've learned and getting you actually using the skills in building Python code.
This course gave great insight in how to approach a new library which I believe is one of the most powerful skills a programmer can have. Keep up the great work that you guys have been doing.
創建者 Matthew A•
This course is a massive disappointment at the end of an otherwise high quality specialization. Unless you really feel that you must get the certificate for the full specialization, your time is better spent elsewhere. Bottom line, this course is bad enough that I will avoid taking anything from this instructor or using the Jupyter programming environment again.
1. The Jupyter programming environment used in this course is awful. The final project (90% of the course grade) takes literally hours to execute after your code is complete. Most of the time, the service will fail and require a restart before the code finishes. It doesn't matter if there are any errors in your code. Jupyter probably will time out before you can actually execute the entire program, even though your program won't be much more than 100 lines long.
2. The instructors could have helped this by choosing a better focus area, but instead, the course focuses on intensive image processing that uses a lot of memory and processing power. I'm not sure why they made this choice, but it doesn't work well for something that is required to be completed in an online environment.
3. This course is poorly designed. One of the main goals of this course, as listed in the introduction, is to teach students to use program documentation to build their own code. Unfortunately, the tesseract and opencv libraries (two of the three main libraries in the course) are not well documented - the instructor says as much during the course - and the student is effectively reliant on the instructor's background knowledge to make them work. I don't know why these specific libraries were chosen, but the experience would be better if they were changed to focus on something that could actually give the student independent experience.
I could continue, but it would be more of the same. I have been very pleased with all of the other Python courses I've taken from the University of Michigan (10 courses in two specializations), but this one is nowhere close to the quality of the others.
創建者 Geon W P•
They just throw you an unorganized super short text book and a video of the prof just reading it without any explanation.
Then they give you an assignment that you can't do with what they taught even if someone actually managed to learn from their material.
on their assignment they literally say "These hints aren't really enough, we should probably generate some more." which just means that they haven't taught you enough for you to do this but oh well
for pillow just go through this site....
it's not worth paying for this course even though the other 4courses in this specialization was good.
This last course just gets me speechless with the way it is made.....
創建者 BO F•
Terrible! There is a huge gap between the first four courses and the final course. Although the first courses are pretty good, the last course is too hard for students to understand.
創建者 TSE M H•
Rushed and Unstructured course, Worst course ever!!
創建者 Jason J D•
This course is a little difficult as compared to the previous courses in this specialization. The instructor does not help you much, but puts out hints and clues at times. Having said that, this is like a typical real world project, where you may have to read through the documentation of various libraries and learn how to use them. Overall, I would say that though this course does not meet the standards of the previous four courses, I personally found it to be useful and challenging! For those who are not aware how to work with complex Python libraries, the assignments could be difficult. Good luck!
In my opinion, this ranks as one of the most abysmal courses I have taken either online or in person. Not because of the course content which is highly interesting and a final assignment which is very challenging but the way the course was conducted, spearheaded by an inept instructor and an apathetic staff support assistant. Plus its disjointed placement as the last course in the Python 3 Programming Specialization
This is the culmination course in the Python3 Programming Specialization and has very little connection to the previous four courses which were taught by instructors who were enthusiastic, empathetic and incredibly good at delivering the lecture content supplemented with great assignments and an excellent interactive textbook. After you complete the first four courses in this specialization and you think you are reaching the finish line, you get sprayed with a blast of shotgun fire in this last course. In this last course we have an instructor whose idea of pedagogy is presenting a set of code and content in Jupyter notebooks and parroting those notes. And you have a staff member who, instead of giving encouragement to students and responding to their questions responds with statistics on how many times a question has been answered and rambles on around the original question without directly answering the question.
The final project is a total disaster and not one worthy of being included in an MOOC. There are multiple bugs in the instructions, no help from any source. This is the type of project that should be given to students attending Go Blue in person with supporting resources such as Grad Assistant tutors and the instructor himself. Instead we are told what packages to import and then do everything ourselves. The first four courses do not prepare you for this poor excuse for a final course.
I have decades of programming experience both as a faculty member and a lifelong student and I was therefore able to complete the programming assignment which is replete with coding and execution problems. The execution times are enormous – anywhere from 15+ minutes for a single run which means that debugging is a nuisance.
In my humble and honest opinion, this course appears to be included in this specialization to enable the course instructor the opportunity to get some enrollment and earn some bucks from the hapless students who have completed the first four courses and feel it incumbent to complete the specialization. If it were a stand-alone course, I doubt he would have many takers. Computer Vision and Recognition are specialized areas. This is a course that needs to be under the umbrella of another specialization. It is not meant for learners who start the specialization with no prior or little programming experience as stated at the time of start of the specialization.
On another matter, when I was peer-grading the assignments, I found that 90+% of the submissions were plagiarized code which I flagged. Out of the 15 assignments I attempted to peer-grade I found only two that had original content. Never, in any online course have I encountered such numbers. This clearly indicates that the students were desperate to complete the specialization and resorted to submitting plagiarized code. The faculty at UM should discuss this more fundamental problem and ask themselves why such a situation is prevalent. No one connected with this course seems to give a damn about this.
This course is a dismal mark to an otherwise elite university and its great faculty and a revelation on how terrible an instructor and his pedagogy can be. It is a travesty to the thousands of students, some from poor countries, who have enrolled in this specialization and paid fees with their hard-earned money.
創建者 Aleksey M•
Compared to the first 4 courses of this specialization, which are excellent, this one is a disaster.
The organization is poor. Installation instructions are full of errata and prone to conflicts in a variety of environments. This leads to frustrated students and teaching staff, who are losing their professional tone in the forums. I get it, I also would not want to be troubleshooting massive multi-library installations for a variety of OSs for however-many-students Coursera hosts. Figuring out "what to do next" requires excessive guesswork and assumption. For instance, simply downloading and moving the lecture's .ipynb files to your local environment is only _alluded to briefly, in an optional video, without ever showing it done on screen_. The course then proceeds into an unrelated lecture. Feels very disconnected and jarring.
The lectures are poor. They consist of the professor talking into the camera with no accompanying materials, and again a lack of clear direction. A concrete example: the professor talks through navigating to /usr/lib/python/site-packages/PIL/Image.py file and opening it in a text editor, all _without showing the screen or terminal_, while explicitly spelling out "See-Dee" (cd) and explaining that's how you change directories... What? There are several problems with this. 1) It is NOT an effective teaching technique to talk through such multiple dir/file traversals without showing the terminal, as it's easy to get lost from where the prof is talking about. 2) Even if there is an assumption here that the students are *SO* comfortable/advanced with *nix already that you can just talk them through that many cd's AND opening a text editor without messing up the file, WHY do you feel the need to explain what "cd" and "ls" do as if it's our first time using it? There is a strong and jarring disconnect here that makes it feel like the course was unfinished - as if the prof believed he'd have an onscreen accompaniment that never materialized. BAD.
The jump in difficulty from the first 4 courses is huge, but it's largely because so many intermediary steps are ignored entirely or poorly explained. The logical composition of this course is nonsense - simple things are explained while complex things are overlooked. If by design it's _intended_ to provoke solution-searching from the student, then at least let the student know that they're expected to find the solution themselves.
Overall I found this really frustrating, and a huge disappointment. Poorly thought-out, poorly executed. I imagine the people actually finishing this course are the ones who are REALLY invested in learning computer vision / these libraries for their own needs, and if you have that motivation, you can give it a try. Otherwise, I see no reason to bang your head against the wall fighting upstream against this course's shortcomings. There are plenty of other resources online for learning python and simpler projects with better documentation, that would be a more natural next step from the previous courses.
Again - this does not reflect on courses 1-4 of the specialization, which are very good.
創建者 Hanno H•
This course ruins the entire Specification. Its incredibly hard to do the exercise, because the lecturer thinks its a good idea to not provide the needed information so that you "learn" by finding the things you expected to be taught online.
The course work is, and I quote, "a project with minimal scaffolding. Expect to use the the discussion forums to gain insights! It’s not cheating to ask others for opinions or perspectives". In other words, part of the learning target is to work through bad or incomplete documentation and to solve an exercise that is stated without providing the information you need to solve it. Unless someone else already went through the moves, the lecturer fully expects you to fail by yourself.
If I wanted to feel the pain of learning without help or instructions, I would have just worked with the online manuals ... which is what this exercise openly tells you they want you to do. Its a shame for the other 4 courses, which are well done and have passionate lecturers.
What a waste of my time.
創建者 Yvonne H•
This course is a complete divergence from the Python for Everybody course and the 4 courses in this specialization. Wasting time digging through pillow etc documentation. Incredibly poor way to start the last course of the Python 3 specialization - unless one is into image manipulation which I am not. The goal for me was to learn python and apply to real world data issues such as forecasting. Modules like numpy would have been much more applicable to work through. I will be taking the data science specialization from another provider (IBM or Stanford (alma mater)) and do not plan to take another course from University of Michigan due to this last course being so poorly administered. The gym analogy was poor excuse for not taking the time to put together more suitable assignments. Such a shame because the first 8 courses were fine.
Absolutely awful course, I would have given 0 stars if possible..
There's an enormous gap between the previous courses and this one; concept are not explained thoroughly and to complete the final project you have to look for suggestions back and forth from the discussion forum or on the internet.
The course really ruined this specialization, which was awesome from course 1 to 4.
創建者 Gannon O•
Abysmal. Just take it down or keep it optional for those that have already completed the other courses. Not only is there little to no actually teaching, the topic is somewhat far removed from what someone might in counter in a real life work setting - All this has been said hundreds of times in the course forums already though....
創建者 Don S•
The project was pretty good. The course content was lacking and the teaching assistant was very rude and unhelpful. I plan to try the next Specialization, but if the first course isn't better I'll be looking elsewhere.
創建者 Sagar M•
Very Poor Course Compared to the previous courses in the Specialization, the teacher is also very monotonous and boring, People should avoid this course if not a part of the specialization.
創建者 Mohammad H•
Very badly taught. Actually he doesn't teach anything. He introduces you to some modules and says go read the rest, and do the project. Disappointing
創建者 Ustinov A•
It's very interesting the assignment and the final project. But maybe you should think of changing the Jupiter notebook. It doesn't work correctly very often.
創建者 Sophie T•
It's not an easy course but it was well worth the effort. I learned a lot about pillow, tesseract and OpenCV.
I give this course 3 stars because: Firstly, I did learn something from this project. Secondly, I am also a little dissapointed with this project.
Let me start with pros: Although my job doesn't require image processing, I believe in the future I may need skills with pillow, tesseract etc. So it is a very good starting point to learn something outside of pandas, numpy, seaborn, etc. Also, this course gives me opportunity to explore different libraries independently with a little guidence.
cons. As this is final project of 4 python programming courses from very beginning to advanced concepts like functions, classes. I was looking forward to a comprehensive project which could cover all the concepts and skills we learned from previous 4 courses, plus a little expoloration of new library. Also, I really love final projects of past 4 courses in this specialization, especially the last two projects (my favorite is the final project of the third course). With milestones, step by step missions, and application of almost all the stuff learned previously.
As I had previous coding experiences in java and C++(not professional programmer though), it is not too tough for me to figure out what to do with this project(still a little challenging in the beginning) I can see how tough it is to those who never coded before.
I do recommend this project to those who have some experiences with python and would love to explore more python libraries. For pure beginners, I recommend UMich to redesign this course.
Anyway, it is a fantastic specialization, and I really learned a lot. The learning platform and in-lecture practice is just amazing. Thank you UMich team. Although I am not alum of Umich, still wanna say: Go Blue!
創建者 M.D N F•
The instructor was just reading the code and at many places, the screen doesn't show what the instructor is intended to show.
Anyways, Thanks to everyone, I learned a lot from this specialisation.
創建者 Ghifari A F•
Not as good as the previous course in the specialization
創建者 Gerson R•
The first 4 courses were like really good, awesome actually. The problem is the 5th one. I think it was a mess due to the following reasons:
-There were uncorrected mistakes that seemed to be there for like several months or even a year without solving.
-The submission system for the final project was a nightmare and the Notepad++ solution they gave us was not working at all (I saw other students complaining about this in the forums). I had to copy my code several lines at a time to ensure that it would not lose formatting and checking that it had the correct indentation at each moment. Even then, after ensuring that everything was alright and seeing my own submission, there were some lines that were not formatted correctly even if I did input them correctly. I had to put a disclaimer to say that it was not my fault if the indentation was not the correct one. This situation has been going on for months and action has not been taken. This is unacceptable.
-We were required to execute someone else's code for review in our own environment. I could only copy and paste another fellow student's code for review when using Microsoft Edge (Chrome and Firefox did not do the trick).
-The staff moderating the forums was not really helpful. I asked about how to download a file that could not be download through the environment, but he would not say that the pinned posts in the forum were intended to help. Normally you do not expect to have that many errors in your course material, so it is normal to NOT expect that many tips to be available in the forums.
-Each iteration of the code is like really low execute. The instructor claims that the big file in the final assignment took like 10 minutes to process, but that is far from reality. This leads to my next point.
-The theoretical contents have not been clear and useful. We were told that certain things could be done to improve processing of the image, but none of it seemed to actually make anything better. For instance, when we applied the bounding boxes to the OCR, it did not seem to actually improve anything. What is more, it made OCR to work worse.
-One of the parameters that were needed to make the pyteseract module correctly was not explained during the course. I would get really bad results over and over until I found one of the parameters and set it to a reasonable value. Even if the course is based on kraken, pyteseract and OpenCV, it is maybe a bit ambitious to expect the student to try and use those parameters, one of them without even mentioning it in the lectures. It would be understandable if the material provided was much faster to compute, but exploring two parameters, each with its own set of values is a really time consuming process.
-It feels like most of the content in the lectures is junk and only 10% of that is really useful in the assigment. The other 90% actually makes things worse (as I said previously some recommended techniques did not work) and is presented in a convoluted way. I had this sensation many times during my university years, were the classes would be totally disconnected from the exam and with real practice. You feel like the lectures have been a total loss of time. I consider this bad teaching practice. I expected more from a university in the USA.
創建者 Samanth N•
The previous 4 courses were so much fun solely due to how the assignments were structured. This assignment however was filled with so many issues along the way. Firstly, computing image data takes a considerable amount of time, throw in face recognition modules like opencv and it becomes even harder. I have a few things I liked and did not like about this particular course:
What I liked:
1) The fact that the week 1 assignment was so intuitive and yet challenging actually kept me hooked on to this course and finish this specialization.
2) There are a lot of real-world applications with the content shared in this course. This happens to be the most (at least for me) practical course in the entire mega course for python.
3) This course offered a fairly good amount of interaction with the Jupyter notebooks.
What I don't like:
1) The week 1 assignment was actually fun and challenging at the same time but, the hints provided for this assignment were little to no help. I'm sure there are better ways to make students read the docs more but, this shouldn't be one of those methods.
2) The week 3 assignment was probably the most annoying assignment of this entire course. The assignment looks fairly easy but, the entire code takes over 20 minutes just to execute. This can be a real issue when you're dealing with experimenting with images and image data.
3) The files for this assignment were not readily available for use. I found out that they had the assets up for download in a separate thread which was not even pinned when it is so important.
4) The notebook was taking so damn long that I decided to install Jupyter and work on my local machine with those files. This was a great improvement over the Coursera Jupyter notebook, though the video guide on how to install and setup Jupyter is outdated.
5) After spending a ridiculous amount of time coding this assignment, when the time came for submission, it was an even bigger pain. The assignment requires you to post the code to a text field but, this text fields removes and destroys all the indenting and formatting. I spent hours on end trying to format it correctly before restoring to uploading my code to online codeshare platforms like GitHub.
6) The review process is even more difficult. You would have to run the peer's code and cross-examine their results. This was a huge pain since running the code can take anywhere from 20 - 40 minutes to complete.
I hope the team takes this feedback constructively and uses the above-mentioned points to help improve upon those aspects.
創建者 Matthew G•
This course was a big disappointment. The subject matter is great but the level of 'instruction' is totally lacking. This course took a big turn away from the rest of the series in that the other used very good pedagogy and this course felt like they spent hours preparing it.
創建者 Lukas F•
not worth the Money, i am a bit disapointed. but still learned much but not because of the course instructions.
創建者 Rohan G•
Compared to previous courses , this course was really difficult to follow.
創建者 Yulia J•
Horrible! Has nothing to do with previous 4 courses in this Specialization.
I would recommend to the instructors to revisit and rewrite the course. It easier for the learners to work on a big problem completing it by small tasks. For example:
1) step #1 : Do this -> Check -> If worked -> move to another tasks
2) step #2: the same logic
.... and so on till the last piece.
Then put all parts together in a working code and submit the final result.
Also, although, having a person supporting the Discussion chat quite regularly is very nice, it would be much productive if the person has a better attitude. Stephen Catto is not listening to people. He answers to people in a very rude way and tries to close the topic right away not really providing a meaningful reply.
To sum up, I will repeat what I said before - absolutely horrible course!