Ask Your Question
3

CNN / DNN Usage question.

asked 2018-08-06 12:21:46 -0500

holger gravatar image

updated 2018-08-06 12:22:15 -0500

So after digging into the whole machine learning stuff(with and without opencv dnn) i want to take some steps back and ask you guys about your opinion.

Is it really worth training / searching for existing model or should you just use the cloud of any provider to do the thing you want? Theres amazon recognition, microsoft mxnet, ibm watson, etc. The more i know - the more i have the feeling that you will never be able to really compete with them as they have:

  • the knowledge
  • the data
  • the infrastructure
  • the manpower
  • the money

I just hate it being dependend on a big company but i also dont like wasting too much time and reinventing wheels. How do you guys see this?

Greetings, Holger

edit retag flag offensive close merge delete

Comments

On the other hand - we the community have

  • the motivation
  • also a lot of knowledge

I guess if you want to compete with big companies - open source should be the answer.

holger gravatar imageholger ( 2018-08-06 12:31:49 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
2

answered 2018-08-07 06:17:51 -0500

updated 2018-08-07 10:31:10 -0500

CNNs are trully remarkable. It is an established fact that they are solving problems that years of previous academic and industrial research could not.

Previously, object classification and segmentation would rely on carefully hand-designed feature maps followed by some decision making process (knns, svms, etc), and a detection system would be as more robust as the feature map was better designed. HOG, SWIFT, Haar, are examples of this.

However, as more and more computational power was becoming available, GPU processing potential was starting to be taken advantadge of and more and more labelled data was available for designing systems, researchers started training a bunch of sequencial convolutional filters (what now people call deep neural networks) to achieve the the best possible feature map to solve a specific problem. And what they realized was that no hand-designed feature map would ever be as good as letting the computer train it.

This was an important advancement in computer vision, however, in my opinion it creates a new problem. Since this strategy is completely dependent on having huge amounts of labelled data, as well as heaps of computational power, what we are observing is that the monopoly of computer vision research is owned by big companies, such as google, facebook, amazon and a few others. And since they can solve their CV problems with this strategy, I don't see any motivation for them to advance CV. Why would they? If they can sell their services, and if the best algorithms are dependent on their data, what is their motivation to allow CV to move to a new and even better era?

For me, I can say that I am a bit disappointed in the direction everything took. Doing CV was the ability to give eyes to a computer, and this was done by trying to figure out what visual features in an object could be used to differentiate them appart from a scene, and find ways to treat an image so that these features would be enhanced. This is over, since no human being can beat a computer in figuring out what is the best way to treat an image in order to solve a classification problem. Now, doing CV is downloading an existing CNN model and doing some tweaks to adapt it to your problem. Its interesting to see that most of recent research papers work on existing CNN models, and expand from there. And no one really understands what is happening there, its trial and error strategy.

To answer your question, if you can find an existing model that works for your problem, you should use it.

edit flag offensive delete link more

Comments

1

Wow - you really got my points!

"...Since this strategy is completely dependent on having huge amounts of labelled data, as well as heaps of computational power, what we are observing is that the monopoly of computer vision research is owned by big companies..." "...For me, I can say that I am a bit disappointed in the direction everything took..." "..And no one really understands what is happening there, its trial and error strategy.."

That's also 100% my impression. Good to hear from someone that's not only in my head :-) Thank you for your well written comment!

holger gravatar imageholger ( 2018-08-07 06:46:24 -0500 )edit

OFF topics :-> The Times They are a Changing and contemplation is a way to solve problem too...

LBerger gravatar imageLBerger ( 2018-08-07 07:28:19 -0500 )edit

I realize that I may sound like a guy that is not able to adapt to a change in the field, but that is not the case. I use whatever the state of the art is. I just don't enjoy doing CV that much anymore, and I fear that because of the state of things, there is no incentive to do different and better. It is really hard for an university, let alone a guy with a computer, to compete with the giants :)

Pedro Batista gravatar imagePedro Batista ( 2018-08-07 07:33:07 -0500 )edit

@LBerger Hehe - good one. Any yes we are living in interesting times :-) May i answer with this

holger gravatar imageholger ( 2018-08-07 07:36:22 -0500 )edit

@Pedro Batista No you are not whining or complaining(i do sometimes) - you have made a good analysis on this topic imho. It only shows me that you actually spend some time on it.

Ok thank you guys for you comments on this - you made my day :-)

holger gravatar imageholger ( 2018-08-07 07:41:18 -0500 )edit
1

@Pedro Batista Don't worry you are not alone with this problem. "It is really hard for an university, let alone a guy with a computer, to compete with the giants". Help people to understand what is deep learning and how to use it to help to solve their problem it's only my task now.

LBerger gravatar imageLBerger ( 2018-08-07 07:47:36 -0500 )edit
Login/Signup to Answer

Question Tools

1 follower

Stats

Asked: 2018-08-06 12:21:46 -0500

Seen: 148 times

Last updated: Aug 07 '18