Gaofen 4: World's most powerful GeoSpy Satellite continues China's leap in Space.

[Tiqiu]

You are looking at things from birds eye point of view while talking about computers, we have moved way ahead in image recognization and with such a super powerfull image captures and super computers like what china has, its 100% possible to do kind of almost real time tracking of objects. 10yrs down the line this was a distant dream, but today many company are doing this kind of tracking using just AWS compute environment.

Fortunately its much more easy to track man made things, with current generation of software tracking the given red car in city with 100s of camera is very much possible. You just need to first identify the car. Chances of false matches will be high if all the cars or majority of the cars are of same shape, size, color and driving in similar patterns. But otherwise tracking things like AC once identified is very simple. And there are various mechanism to split and identify. With super computers like what china has its very much possible to track not one but 100s of vessels, with relevant data. Probably they cannot do it on day 1 but in short period of time when they have ample data to do mining and mapping things it would be very much possible.

Similar techniques are used at some ports for tracking movements of containers using highly placed cameras and this helps in developing plans to optimize the entire process.

A insightful read and opened a new angle for me.

 

[Viny]

You, along with other Chinese member here, have misunderstood the meaning of "Track" and the Meaning of "MOnitor"

It's very important for you to understand, to track something, you need to recognize something first. Only you can Identify something over the general area you are monitoring, only then you can start tracking that thing.

Now, imagine this, you are in an ocean, to look for a target, you first need to zoom out to maximum resolution so you can cover the biggest area and take in all the data into your feed. Then you start looking. And consider the optical physics, the further you look at an item, the less detail it can get. And what we are discussing happen in this stage. Alternatively, How do you recognize that is an aircraft carrier when you are looking from afar.

For a human eye, you recognize shape as they appear in curve and contrast, a human face is always paler/lighter than their hair. So you can recognize a human face without any feature of their eyes, nose and mouth. However, it does notwork as a object, because the feature of an object is flat, and uncontrasted. How do you decide whether or not if that is an aircraft carrier or an oil tanker by looking at its image 1000 miles away from you? Given if we can zoom that far? You can't, that's pysically impossible to do it.

For a computer, it would be worse, because a computer only recognize a pixel, and for them a pixel itself means nothing. Unless you have defined a set of instruction to the computer what that is with that bunch of pixel. Now consider this, for a computer, they recognize not the shape of an object, but a change in bits or bytes thata represent a single pixel. But for them, it mean nothing, for a computer without instruction, a change of bits/byte is just that, a change from FF0000FF to FF9999FF.

Now, if you start to put instruction to a computer, then they can interpret what's that change is, however, how do you instruct a computer what is an aircraft carrier? That is what we are talking about here. How do you define an aircraft carrier to a computer? Because without that, a computer cannot understand what is an aircraft carrier..And thus, this is a moot point, because the computer will simply see something on top of something and may be displacing, and that would be the best interpretation any supercomputer can show you.

In Satellie works, the computer canonyl highlight the "Anomaly" on its feed, hence the change of pixel from one square to another. but to determine what that change is by itself? That is some kind of super algorithm I never heard of.

Firstly tracking is only initated when you have some identified object. If you dont know what to track you cannot.
So if you are trying to search and track for a given AC then you need to know what AC and in which area you want to track it. When tracking an object other than images there are 100s of different sources of information that goes in when you want to track something. So its not out of blue, also before you think of tracking something a big data set is build about the object to track and how to identify it.

Coming back to your idea of how computer recognizes images is as at best of laymen, probably thats how it was initially thought of and started with. Current era is very different there are 100s of algorithms and methods in which image is processed to find near matches and not specifically doing it by bit/byte matches or pixel matches. Binary match is not the way how image recognition works now.

Visual definition of an object in computer is very different then what you are thinking, we are moving towards algo matches that can even beat how eyes and brain interpret things at time. Stock photography industry is prime example how there visual searches are doing great, infact now you even have search engines where you can upload an image and it will search for image which are similar, for example if you upload and image of leaning tower of pisa it will find similar match from globe. Try https://www.tineye.com its just a very simple consumer oriented implementation.

 

[Dungeness]

You got a negative rating because you used the word "loser"? That's not innovative even for someone with a clear agenda because it clearly shows the abject hatred.

@Hu Songshan, @waz , @WAJsal , I remember the same person who gave a negative rating to Dungeness because he used the word "loser" insulted a member's mom, saying something like "your mom is my dog." There is still a link to the post.

Why nobody bothered to sanction the said member to openly insult a member's mom but it is perfectly OK to receive negative rating because one uses the word "loser."?

So you think "loser" is an insult? If so, how many more million negative ratings should have been given so far?

I guess one needs to take the issues of fairness and equal treatment seriously. I hope that you would, this time, solve this issue and not going perfectly soft about the said member.

Regards.

Knowing his person's demeanor and in the past against Chinese posters, I did not even mention any name, and my post started with "IF this is indeed his agenda.......". How dose this particular post violate any rule of PDF? I am wondering just exactly what makes this person so special on PDF? Is he representing some interest group that PDF can't do anything about? @Hu Songshan @WAJsal

 

[jhungary]

Firstly tracking is only initated when you have some identified object. If you dont know what to track you cannot cannot.
So if you are trying to search and track for a given AC then you need to know what AC and in which area you want to track it. When tracking an object other than images there are 100s of different sources of information that goes in when you want to track something. So its not out of blue, also before you think of tracking something a big data set is build about the object to track and how to identify it.

Coming back to your idea of how computer recognizes images is as at best of laymen, probably thats how it was initially thought of and started with. Current era is very different there are 100s of algorithms and methods in which image is processed to find near matches and not specifically doing it by bit/byte matches or pixel matches. Binary match is not the way how image recognition works now.

Visual definition of an object in computer is very different then what you are thinking, we are moving towards algo matches that can even beat how eyes and brain interpret things at time. Stock photography industry is prime example how there visual searches are doing great, infact now you even have search engines where you can upload an image and it will search for image which are similar, for example if you upload and image of leaning tower of pisa it will find similar match from globe. Try https://www.tineye.com its just a very simple consumer oriented implementation.

As I said, as far as I concern, today technology does not goes into that advance, I cannot speak for the term in the future, or some sort of alien tech whether or not China may have processed. Today, it is impossible

Tineye works by compare the exact image you supplies and look thru internet and find that exact image matching those datapoint. It does not work as a smart recognition (A simple pipe and search program, it took some data point and look thru the database of image over the internet and find you a match). And I am talking about how do you recognize something in multiple angle. And how computer can determine that is something but not other.

TinEye does not typically find similar images (i.e., a different image with the same subject matter); it finds exact matches including those that have been cropped, edited or resized.

Yeah, same here

??????????What the hell are you talking about??
Is there a point in your post that worth replying?? Honestly, I don't understand what your post mean?

You want me to comment on your water bill or the fact that it was usually the state government ombudsman deal with these Utilities and it have nothing to do with the federal government? Unless of course you are accusing the state government employee of being unfair or corrupt.

 

[Viny]

As I said, as far as I concern, today technology does not goes into that advance, I cannot speak for the term in the future, or some sort of alien tech whether or not China may have processed. Today, it is impossible

Tineye works by compare the exact image you supplies and look thru internet and find that exact image matching those datapoint. It does not work as a smart recognition (A simple pipe and search program, it took some data point and look thru the database of image over the internet and find you a match). And I am talking about how do you recognize something in multiple angle. And how computer can determine that is something but not other.





??????????What the hell are you talking about??
Is there a point in your post that worth replying?? Honestly, I don't understand what your post mean?

You want me to comment on your water bill or the fact that it was usually the state government ombudsman deal with these Utilities and it have nothing to do with the federal government? Unless of course you are accusing the state government employee of being unfair or corrupt.

How about trying this

download this image ( a typical leaning tower of pisa image).

Go to http://images.google.com
In search box, click on camera icon (search by image option), select upload image file and then select this image to upload.
Then click on search.

It will give you two results. First exact same image and then second option as "Visual Similar image"
Click on "Visually similar image", now this will give you 100s of other images of leaning tower of pisa, which are not of same angle, size or color or composition still matched.

And this is part of technology which is publicly available and opensource version of same is also available for buiding similar visual search solutions. So you can probably think of advancements that advance research labs have in this field.

So this is no alien tech, if it was so much of alien tech then driverless or autonomous cars would not have been running on roads. They also do a lot of visual pattern recognition at real time.

 

[Tiqiu]

Firstly tracking is only initated when you have some identified object. If you dont know what to track you cannot.
So if you are trying to search and track for a given AC then you need to know what AC and in which area you want to track it. When tracking an object other than images there are 100s of different sources of information that goes in when you want to track something. So its not out of blue, also before you think of tracking something a big data set is build about the object to track and how to identify it.

Coming back to your idea of how computer recognizes images is as at best of laymen, probably thats how it was initially thought of and started with. Current era is very different there are 100s of algorithms and methods in which image is processed to find near matches and not specifically doing it by bit/byte matches or pixel matches. Binary match is not the way how image recognition works now.

Visual definition of an object in computer is very different then what you are thinking, we are moving towards algo matches that can even beat how eyes and brain interpret things at time. Stock photography industry is prime example how there visual searches are doing great, infact now you even have search engines where you can upload an image and it will search for image which are similar, for example if you upload and image of leaning tower of pisa it will find similar match from globe. Try https://www.tineye.com its just a very simple consumer oriented implementation.

I am sure India will gradually develop this type of capability of tracking down the road as well. The advantage of having AC is not only you can learn how to use it, you can also study how to detect it from under sea and space.

Leave it, in fact you ain't worth my breath period

Brother, please leave it to the Mod to deal before getting yourself into another trouble. Not worth it

 

[WAJsal]

Please do not derail this thread, please do not resort to cheap attacks. Consistently going off-topic and derailing a thread will lead to a ban.
A complain and protest can be lodged in GHQ section, you can take your complains against a negative rating here:
https://defence.pk/forums/general-headquarters.170/

And if you have a complain regarding a member and any abuse of power, you can complain in GHQ. Chinese section is too hectic, it's not the most perfect forum but we try our best. Please play your part and help us keep this place clean.
If you see a troll posts, please report it. I will myself try to deal with that report and make sure necessary action is taken.
Please try to remain on topic and the topic is China's leap in Space, discussing other members has nothing to do with topic on hand.
@TaiShang ,@Dungeness ,@rcrmj ,@Tiqiu ...

@Hu Songshan , please open a PM with a number of Chinese members complaining, look into their complains seriously and see how you can help your fellow countrymen.

No more off-topic posts and no more complains here please!
regards