I am a first-year doctoral student in Computer Vision, and have some confusion about my research direction recently. I hope I can get some help and advice there.

A couple of months ago, I was arranged to do research about optical flow, which is a kind of motion representation and can be used in many computer vision tasks. In these days, I read many papers and did lots of experiment about optical flow about its history, datasets and criterion. Therefore, I began to believe maybe I could make some contribution to that. Since some previous ideas are proved not working, the professor suggested me to give up the research on optical flow and change to another.

Now I am very hesitant about the next plan, there are two reasons. On the one hand, I am not content to give up just like that, and I think we could do more. On the other hand, I know very little other directions in computer vision, so I am not certain which direction I should choose. For computer vision, I want to try everything. Although optical flow is not my favorite one, I can insist on it. I really want to choose a research direction, and spend several years or the whole PhD stage to do some meaningful works.

What should I do? I hope experienced people can give me advice. Thank you!

  • 5
    This is a discussion you should have with your advisor(s)...
    – Solar Mike
    Mar 15 '19 at 10:07

You should be talking to your advisor(s) and you may have some good points to discuss.

If you think you have some relevant ideas, then you should discuss them with your advisor - one may be the "key" you are looking for.

If, after considering your points, the advisor still has good reason to think that there is no future, then you should accept that decision, clear in the knowledge that you put your points across.

So, discuss with your advisor and make an informed choice with them.

  • Thank you, I think so. Maybe I should more assertive. Mar 15 '19 at 11:39
  • 2
    It's not about assertive. How well can you trust your instincts? Do you have experience in it? Your supervisor can be saving you 3-4 years of your life by suggesting a switch of topics. Of course, you may be right and you should be pursuing optical flow. Do you have experience in the past where you went on your own and were scientifically successful? If so, give it more time. If not, perhaps give your advisor more consideration. Mar 15 '19 at 15:30

Some people have enough ideas and a secure enough background to be able to carry on doctoral studies with very little advice or direction and be successful. Others depend more on their advisors, who usually have more experience and insight into what is likely to work and what is probably unproductive.

You need to decide whether you are more like one of these or the other. Often the professional insight needed to be successful in research doesn't come in the doctoral study years, but within (hopefully) a few years afterwards.

I worry about your statement that "I know very little other directions in computer vision", which makes me wonder if you are ready to take the lead at this moment. If not, then you might follow your advisor's advice at this time and wait until later to explore the wider world. But you need to decide for yourself whether the possibly more risky path is right for you.

People can be successful either way, but not always the same people.


It depends how convinced you are and how willing to go against accepted opinion. History is full of predictions of impossibility, for example the advancement of AI in general and neural networks in particular nearly came to a dead stop at one point because of a book written by influential authors.

In 1969 Minsky wrote the book Perceptrons (with Seymour Papert), which became the foundational work in the analysis of artificial neural networks. This book is the center of a long-standing controversy in the study of artificial intelligence ... It is claimed that pessimistic predictions made by the authors were responsible for an erroneous change in the direction of research in AI, concentrating efforts on so-called “symbolic” systems, and contributing to the so-called AI winter, a period of reduced funding and interest in artificial intelligence research. This decision, supposedly, proved to be unfortunate in the 1980s, when new discoveries showed that the prognostics in the book were wrong. http://scihi.org/marvin-minsky-artificial-neural-networks/

I did a Masters back then and wanted to do my dissertation on neural networks. I was strongly advised not to because 'the subject is dead'. I went ahead anyway, got my Distinction and sure enough a few years later neural networks were back in fashion. My paltry efforts were soon overtaken by cleverer people but by then I had moved on to something else.

If you are very confident of your abilities and are willing to gamble somewhat then you may prove everyone else wrong - it's happened many times in history, just read a few biographies. Are you tough enough?

  • Thank you, it is full of chance and challenge, and I would think about it carefully. Mar 15 '19 at 11:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.