It will give you all the cars And you would have seen this kind of stuff on you Pube and Lyndon off the people posting people are walking and your Lawrence’s dear able to identify them on in the image Yeah orders You can see and see houses You can see a pole You can see a building A lot of things you can see Yeah Sometimes our application does not need this kind of sophisticated I will give you an example Let us see You were having um again coming back to your parking slot The problem I think the Williams No don’t say that At some parking lot which has since It’s not since those putting a sensor that very costing something People maintenance a lot of issues Maintenance carnival owns.
What if I remove the sensor and I put a simple camera with it and this camera is connected to one of four Um deep learning Nichols Yeah Now what It doesn’t The camera captures an image of this law now is it necessary for me to let s say there’s there is a car park now the only intention to show years whether this car park is full or indeed has to detect and displayed in you know in a display board down the line so that nobody comes in ways that time of year And yeah So what is our job here is just to identify Is there some object on this What is the object What color What company were the type of car were not interesting applications They don’t need to detect like this But yes you need to separate the object by color or by any of the other makers.
We’re going to see segmentation will beans So this is called semantic segment is Then you are trying to give a color Two similar type of objects People see that all the cars are almost probability Yeah well you know I almost did kind one color-coded of wonder And because of the little one on the ground of the ones Now try to bring that you know the including decoding concept of what is gonna happen now we will make an Al Gore them so chilly that if this color is red if it is incredible stuff what I would do is I would see toward the neighboring stuff Neighbouring excess to this also get the same color But in this case our segmentation is getting purple color.
Yeah, so I will see if this purple purple purple purple purple I’ll try to give purple ever almost everywhere till I find it remains in right And also you will see there are some borders you might find the poker color is coming apart from So this is the way you can say that Yes the image could be and quartered down And then when they’re decoding back there are chances that we can get a very similar color around a simple object Okay so this is our application off including vehicle Now you may actually find apart from this parking lot of a does need I notice that you have You know how you are walking on this little car Self-driving car So what does a car need to see.
A car has to do things the hard drive You know very well it right Obviously has to find a hole Yeah No let us say that they may go into the car There could be a lot of countries and there would be one road going over here The car will not come People Sometimes because of Lord objects present the name you’re not sure So what I would do for that time is to solve this problem I’ve been seeing I’m putting on garden saying that wherever I find a road, first of all, we will see any made like this And whenever there is a road you might find one particular color.
I don’t care about any other object around I only care about one thing with his little That’s yeah So there are algorithms in which we can black and this so whatever you see here But if I say ok I’m focusing only on the old apartment I can I invite this out into one Come let us say whatever is not roll is okay Something like this Image a picture then This is happening Not only do colors when Israel and one are black the particular self-driving car will come to know that Yes that solves my road However I keep findings black color I will keep falling And there could be one more application on that If there’s something on the rule it has to stop and manage the speed, In that case, you can say that Yes I have detected a person is protected a car like this Awesome It should not be that much complicated as competitive driving.
Yeah So this is one example of Semantics signatures Another various type of cemented segmentation We have instruments and we help someone So let’s start to Listen to Mindy because I see it if you obsidian that are two buses, Okay So when the image was shrinking down and when we get back the image what’s gonna happen is then the pixels are going toe You know basically say that I get is the part of the same family letters use the Same color So I let us in this world my first Nixon one minute up This was my foot My first pizza.
Then I would have brought a bracket around it More Same color pixels More more more like that I started increasing and I find the board whenever that is bus EUobserver then the green color Wherever the bus is no bus it is followed by you I can easily come to know that when I take a look at the two buses that are So this is quite Symone Big segmentation and then some people Instant segmentation Now in this also if you want classifications because this is one type of west this is another in that gifts the incident with the double-decker And this is okay so all we need to do it We need to have a very good corpus That is your training on data And we need to have your started getting Dana Are you really getting We have to train them Just a phone saying that if this is the school you have to do this Um only if this is the school.
You have to do this wherever you find this kind of buses Eucalyptus I need to have original images And I couldn’t do have to have all in anything else possible him apart from these to see if in case you trained in it saying that I just want to take this I don’t want to predict this double-dip that came This also become black This would be okay So once you progress under that you can have Lord of the application Any questions Raise.
Now it is okay Oh how does this coloring up and off The corpus already has this column for Don’t have also in court It’s not city No The question is quality has the color The carcass has both days *** It has the original image and it has bordered on good England accordingly Yeah class We’ve been using unit onto this agreement All right True Down the line in the case that’ll work in a mosque This is must These images are called mosque These images are so we Morehead So So this is what I was trying to explain but Okay we would see this in the unit I don’t want guns You guys carry Matt the morning One more person go back So that’s Yeah How do these start Initially.
I’m taking the anchor point ascent of It’s a brand new do that The segmentation on Yes Before that I find with Google unit boats we look at the unit structure Whatever you’re doing now is a few people Remember If I ask you one simple question if this is my original image and this is my convenient good image what can you say about these two What is present in the until Yeutter image and somebody did means What does convolutions feature map It shows the battle Can I say the most important element Mostly under Don’t Yeah you made a that Actually the independent media is broken down into different So we’ll have a list of printers available It’s really And what are these They’re one of the most important features.
What a friend Thank you So just imagine that an image is broken down into a very simple and quarter part So just imagine this is a small list but I can say a combination of Lord of a new district So can I say these out Is that one of the most important features for me Remaining stuff We don’t want much Get about it Thanks Next What I do is if you observe now we are up skin So we’re picking up each breeder and as I surely people that if this is one for if you observe the signs of this inside of this definitely didn’t include this So by fighting what are we doing If you remember Max Tooley Maxwell he wanted it out off out of this four Well because the maximum speech we do reverse office even see that instead off that Max pulling it We will replicate this by the same pizza So the pizza number is one That meeting we present.