 Very good afternoon to all the educators who are joining us. Very good afternoon Rishi. We'll be just shortly starting in a moment. Very good afternoon Priyanshi, Pooja Ishwari Shiv. Now we can see some good chats here. Very good afternoon to all the educators who are joining us. We'll be just starting shortly. We are just waiting for other educators to join us. Good afternoon Yukti, Vikram, Christina, Yukti and Gauri. All the educators. I'd just like to take a question. It is mentioned like Sana has mentioned. If there is some problem, telegram cannot be accessed at Pakistan. So not an issue. If you want to reach out to us and you are not able to join the telegram group, you can also reach out, send your queries to support at the stempida.com. As at the last, we generally show you our contact. So we'll just write the email ID so that if you are finding any problem and you are not able to join us on the group, so you can reach out to us on the support at the stempida.com. I've shared the email ID. Please feel free to write us. So Mr. Vipul is also with us on the chat. So if you are finding any problem or if you want to ask anything, feel free to put it on the chat. Very good afternoon to all the educators joining us from around the world. It's wonderful to see you back again on the session 3. We'll be just starting shortly in few minutes. We are just waiting for other joiners to join. Are you sure you can just give us a short recap of last session maybe? Sure. I'll just stop my share and re-share the screen in a moment. Am I audible? Yes, yes, you are. Let me just check. I think my screen is a little stuck right now. Okay, so is my screen visible? No, not yet. Okay, sorry. You should be rejoining the session. So let's wait for a second. Yeah, he is rejoined now. So good afternoon, everyone. We are very excited to start the third session. We will be starting out very soon. So within like three to four minutes, we will be starting the session. We are expecting more attendees to join in and then we will be starting. So I'll just re-share my screen and we'll just take a two minute of recap of what we did yesterday. So is my screen visible now? Yeah, it's loading. It's still loading? Yes, yes. I'll just stop. I'll just re-join in the session. Just give me a moment. Yeah, so we were having some issues in that particular room. So I have rebooted the room so that all the participants can join in again. So, yeah, so we can start with the session and we can continue. Yes, Ayush. So we'll just start with the session of a recap. Okay, first we do a short recap. I'll just re-share my screen. Is my screen visible now? Yes. Okay, so I'll just change. So this is basically a session two, a small testing. So I'll just change it to like, yeah. Okay, so yesterday we just saw first we saw the wonderful projects and then we started with the object detection. So in object detection, we understood that how exactly it is working and what exactly it is. So here we just saw that it's a collection of tasks performed under the computer vision where it not only just finds the objects, not only just locates the object, but brings the bounding box and also just put up a tag there. Now, we also saw that how exactly it is going to work. So it first just segments the input, then bring up small little boxes over, like all these small boxes all over the images. And then at final, combining those small boxes after validating the different objects, it's bring up a final bounding blocks around the, I should say, around the objects detected. Okay, then we made an activity where we learned some of the concepts of cloning, we learned some of the concepts of resizing the clone, also like bringing up the counter, right? And we were able to basically find the different object, yeah. I think your screen is not visible. Can you just share it again? Okay. Is my video on audio is clear? Yeah, your audio is coming, but your screen is not visible. Okay, screen is not visible. So I'll just try to reshare again if it is not, if it won't be working, I will just request you to share and I'll just give them a recap and everything, okay? Yeah, yeah. So is it visible now? No, now it's coming, but let us see if our participants are able to view. I am able to view it, but let us see if participants are able. No, they are not able to see your screen. Yeah, now they are able to see. Okay, okay. So if you are able to see the screen clearly, please write it in this chat. Yeah, now they are able to see. So yeah, continue. So I'll just take them from here. So where we discussed about what is object detection? How exactly it is working? Then in how does that work? We just understood about the segmentation, different bounding of objects than combining and overlapping of different bounding blocks and bringing a single bounding box on the different objects. Then we made an activity of an object detector where we just detected the three objects in an image, where we detected the bus, two persons, using some of the wonderful concepts of morning object detection and also adding some good extensions of artificial intelligence. So I won't take it more now. Okay, because I would just because we will start with today's session. Although if you will be having any doubt, we would love to answer all your questions towards the end. So I'll just stop my share. Yeah, I'm just reading my screen. So once again, a wonderful good afternoon to all the joiner, all the educators around the world. We welcome you to the artificial intelligence bootcamp session number three. And it's amazing to see you all on time and it's great. And we'll just we'll now start. So I'll just request Mr. Pankaj to just share some light on what we are going to do today. Yeah, thank you, Ayush. It's very good to see all the educators coming back to the third session. So today's session is very exciting. We will be learning about machine learning, machine works and like how you can teach machine to like learn something and then test it out in crypto blocks as well. So today is a very exciting session where you will be learning about a technique using which you can create a lot of variety of projects. Very versatile technique and it can be used in a lot of manner as well. So I would just begin with the first thing is if you have joined our telegram group, then it's very good. You will be getting updates. Your doubts will be solved there. All the things are happening there. But if you haven't joined it yet, then you can join in using this link. You have the link in the chat as well. If you are facing any problem in getting into the telegram group, there will be two reasons. One, you might not have the telegram installed on your device. So you have to install it. And second, if you have the telegram installed, then you have to sign in into it as well. So after these two steps, you will be able to join into the telegram group. So I am just putting the link onto the chat so that you are able to get the updates regularly. And the second one is we will be doing the activities in PictoBlocks. So if you haven't installed PictoBlocks, then you can install PictoBlocks as well. I would just run a small pool side by side asking all of you, have you downloaded PictoBlocks or not. So it will help us understand how many of you have PictoBlocks installed. And if you are getting any issues, then write it in the chat and we will be able to support you as well. Now, the next important thing that I wanted to convey to all of you is we have a very good community of PictoBlocks users, educators. And we have the community. The community is actually contributing a lot in creating a lot of projects, creating new projects as well. So I am just putting the link into the chat box and you would be able to see the link. And I would request all of you to join our PictoBlocks community on Facebook as well. There you will get an update from other educators as well. It's a growing community and we have more than 1800 educators already part of it and they put up new and new videos and new and new concepts to work every week as well. So it's a community of PictoBlocks users and we discuss all most of the things there about PictoBlocks, about its update and its other things as well. So please join the PictoBlocks community. Ayush has put down the link into the chat and if you haven't joined it, then you please join it. Now coming to the assignments that were submitted by a lot of educators. So the last time, this time also a lot of educators have submitted a very good assignment. They have been testing the object detection. They have been remixing the project and creating something new as well. So I would just want to show some of the projects that were very innovative and we wanted to share some of the projects with you as well. So I'm just starting the video. So this was a very exciting project I wanted to show. I would show the next project to you. Very creative. So in the last session, we just told you about the concept and you have created a lot through it as well. I would just play the last video for this. Hello everyone. My name is Nihagar and I'm working on Object Detector. Just learn in AI Educator Bootcamp. In this project, a girl is there who will tell about the objects present in the room. It is really very fun to learn PictoBlocks and that too in a so easy manner. Thank you for providing this facility and giving us this opportunity to learn PictoBlocks. Thank you. So it's very good to see the amount of creativity coming out of the educators and the hard work that you are putting in order to do the assignments in order to get hold of the concepts and do the activity as well. So this thing is something which is very motivating for us as well and we would be looking forward to see more and more assignments and the videos about the concepts by you as well. So a very, very good thing that you are very much engaged into the assignments and doing the assignments as well. Now, coming to this part again, so I would just remind all of you about the prizes and the states that we have for you. So as you know, you will be getting a digital certificate and the batch as a Sampedia Certified Artificial Intelligence Educator when you complete all the assignments of this bootcamp. So there will be four assignments, two of them are already being told to you and today will be the third assignment and then tomorrow will be the fourth assignment. So for all the PSI deadline is 2nd of July but we want you to submit the assignments within 24 hours so that you can focus on the next activity. So we have been receiving a lot of assignments from a lot of educators but it's very important for you to understand that if you are submitting the assignments and if you are submitting all four assignments which are like proper assignment, not like where you are creating a video and showing us how you have like showing us the output then and then only you would be getting the certificate and batches. So it's very important that you put the hard work to practice the concepts and make the project and then record it to submit it as an assignment. So it's very important that you do it and submit the assignment to us and the educators who complete all the assignments would get the digital certificate and the batch of Sampedia Certified Artificial Intelligence Educator. Those educators would also get the access of the AI resources and the curriculum for your classroom and top seven educators will get Artificial Intelligence Kit which is Quarky Innovator Kit. At the end of this session I would also like show you some of the demo projects of Quarky so that you are more excited about it. So this is something which is at stake and we want like all of you to be very sincere in terms of submitting the assignments on a regular basis. So it's very important that you go over the assignment and submit it in a proper form. So whenever you are submitting the assignment if you are submitting it via YouTube then you have to make sure the link is proper and you have given us proper access as well. The video should be unlisted or it should be public. It should not be private because private videos we cannot see and we cannot judge your assignment as well. Then the second one is if you are submitting it in from Google Drive then please give us the public link of the video so that we are able to access it as well. So yeah, so like second of July you would have the day off so at that particular day you can like it will be very good like if you haven't submitted it yet then that particular day you can use it to practice and submit the assignment as well. So I would request Ayush to get back and start the session of like start today's session. So today you will be learning about machine learning and the concept of how you can like make machine learn something and then use that machine learning model in vector blocks as well. So Ayush now you can take it forward. Yeah, so thank you Mr. Pankaj for just giving us so much brief about all the things and also I would like to tell like it is really very great and awesome to see the wonderful projects by the educated. So some of you have given your best efforts and it is amazing to see. So some of you have used the creativity, some of you have just practiced the same but some of you have also tried to make it personalized. So some of you have just tried to put the object detection and the celebrity detections with your own images. It's really wonderful to see that and we will just love to get your all assignments on daily and as told by Pankaj like it will be we are giving you the time till 2 but we don't want you to like pile the things up. So start submitting it daily. Although in your emails and even in the telegrams group you will be able to see the messages where there will be links separately for different sessions. So make sure you are submitting your correct assignment on the correct submission form. So now I would take you towards the journey of today and that is like a machine learning. So I'll just share my screen. So I hope my screen is visible. Yes. Wonderful. Also like if in case you find any trouble please let us know about in the chat so that like the things can wait to us and we can just like check it down some of the things. So now you can see a machine learning. So there are basic words in it. First is machine and the second is learning. So as the word you will be able to understand some of the things it is something where we are making our machines learn. Now the things come up that how we are doing it and how exactly it is happening in back side or it should say in back end. I'll just take you through these concepts. So machine learning focuses on the applications that learn from experience and improve their decision making or predictive accuracy over the time. Also machine learning is a system where the machines are learning the things by themselves. Now talking about artificial intelligence and machine learning. So machine learning is basically a subset of artificial intelligence. It's not like a different topic but it's in a subset of artificial intelligence. So you can see here in the picture as well like we have a big globe of artificial intelligence. So this big globe contains many other techniques and many other concepts and one of the best concepts is machine learning. Now to understand this machine learning in little more depth, I would first like to give you a little idea of the human learning process. So in human learning process basically there are four major steps. The first step is since the environment. Second is analyze the information. Third is decide and act to increase the knowledge. So if I want to give you a short example here. So if I'll take you towards your childhood so you will be remembering one thing and even we see with the kids also that the first time we started acquiring the knowledge was something from a picture book. Already we had a picture book and there were the pictures of different animals, different vegetables and such all. So there we were able to we used to see it and daily like day up day down we used to see it and we used to grasp the things from there since the environment. This can be the dog. This is the cat. Also that picture book contains the label. So it was not that it just only contains the pictures but it was mentioned there that this is something, a picture of a cat. After that the second part comes up where it is analyzing the information. So when we grew up a little bit we started going out with our parents, guardians, teachers. So we started seeing the real animals. So we'll just take an example of cat and a dog. So we started seeing the real cat and dogs and there we started just I should say relating the real cats and dogs with the cats and dog we saw in the picture. Although the color or the texture of the skins might not be same but the other features like eyes, the tail, the nose, the face and the body structure was the same and we were able to analyze the information. Not only this we also analyzed some of the particular I should say features of those animals like cat can climb the walls but dog can't. Dog are sometimes very dangerous. So we started analyzing the more information about them. And after that the third thing which was a position where we decided and acted upon the analyzed information by us. So just take an example. So it happened to me and definitely generally it happens to every child who is growing. So what happened? I just thought of going to my friend's home and I just took my bicycle and I just started riding it and going to my friend's home and there in between I just got a group of dogs. So what was my decision at that time? Do I decided to just pass them off or go to my friend's home? No, because I already had information where I knew that these dogs can put me in danger. These dogs can chase me. So what I thought I just thought of coming back to my home. So this was the decision making ability, action taking ability which came up after analyzing the information. Now after that what happened? The next day I went to my friend's home and just I entered the home and I was able to find a dog again. Although it was a pet but for me as a kid it was really again a surprising that I just did not able to came yesterday due to the dogs but today again there is a dog but my friend told me this is a pet dog. You can play it around. You can just have a fun with the dog and here comes the part that is increasing the knowledge. So now with these understanding we start increasing our knowledge and we are able to take the decisions. So if you will see here that third and fourth step is something which carries on. So you decide or take action on anything, any problem or anything you do there is a certain output. So it can be corrected, it can be incorrect. Whether it's a correct or it's incorrect this is going to get inside your knowledge base and this will be benefiting you, increasing your knowledge and taking the decisions in the further opportunities. Now coming to the part that how exactly we have developed this knowledge base. So we have developed this knowledge base with the use of senses. So we have different senses like sight, hearing, taste, touch and smell. So now if I give you an example that if you are sitting at home inside your room and you just hear a barking of an animal I won't say the name. The barking of an animal if you are sitting inside your home after hearing that you don't need to go out and say this is a dog or what animal. You will be able to just say at that moment this is a dog because you have seen certain environments you have heard certain voices with your hearing part. If I will give you a very good example again if you are going to any bakery or if you are going to any sweet shop you don't need to taste all the sweets or all the things there. Just anyone saying if I just take an example this is a pineapple pastry you don't need to taste every time. You have already tasted it before and the taste is in your mind and you are able to take the decision do I need it? Yeah I can of course. Similarly with touch and smell the things because we have acquired the knowledge and we are using it. Now talking about the machine learning it is machine learning is also a similar process the only thing is these things which we used to do are carried out by the machines although definitely we have to help them but everything is taken care of by themselves only. So you can see here there are three different stages. So here first is the input. So as when we were very when we were child or we should say when we were very young so we saw the pictures right? So again when we are doing the things with machines so we have to give an input with a certain label. So there should be certain input mentioned with the label the sound that can be audio, video and different inputs. After that the second part comes of a model. So model is a place where algorithms comes into play there is a calculation so there are many different algorithms which take part in the different machine learning concepts and they come up with a certain calculations. And after calculating those images those inputs they give us an output. So this is a small machine learning I should say exactly how is it working but there like I would like to tell you the machine learning cycle. So basically it involves four steps. First is select and prepare the training data set so you can see on the first image so we have a data set so we have created it although we will be working with these all things in coming time. The second part is training with the data model. So as I said model is a place or you can say model is a set of calculations and then after creating it we have to test. So here on the center image you will be seeing two globes here where it is mentioned like training algorithm using training data calculate error using testing data. So here you will be able to see two arrows so this is because we train something we just check it we test that if we are satisfied then we move on but if we are not satisfied what do we do? We again train them and this is a process like as I mentioned it's similar to a human learning that if we are not good at something we keep doing it till the time we become the master and at the last of course we have to deploy it so for this deployment definitely we are going to use our own tool on which we are working so from like two to four days that is pictoblocks. So now I'll just stop here and I'll just like to take a small poll with all of you. So is the things clear till this? Like is the things are very clear till this if you have any questions you can put it on the chat. Yeah hello. Yes Ayush 90% people are clear so I think most of them have understood the concept and in order to just summarize it I would summarize in 10 seconds like 10 to 20 seconds so basically like if you have to teach yourself about like knowing about two new species okay if you haven't seen dinosaur and you have seen two species of dinosaur then what you can do is you can just look at the photos of 100 photos of one species 100 photos of two species and then after seeing these photos if a new photo is coming up then you can identify the dinosaur so that's the power of our brain and the human learning process and this is something which we will be doing with the computer and the machines as well. Yes Ayush you can go ahead. Okay yeah so like I think yeah of course this is a wonderful example educators so even we can try this okay it's not just a saying but you can try this like you if you haven't seen different I should say vehicles also right so what do we do is we started searching for that vehicle and after seeing 7 to 8 or 10 or like I should say 100 of images if we are seeing even from the backside or a side that on the road we are able to just say without even seeing the name right so okay this is that particular car so this is the power of basic basically the brain which has been told by Mr. Pankaj so now I will take you to the picture blocks and where we will be doing this wonderful activity of making a mask detector so I will just share my screen of picture block so I will just check with Pankaj so is my picture block screen visible yes yes it's visible okay wonderful so like educators it's been now the third day that we have been doing the things in picture blocks and we have seen your love towards the picture blocks and doing and putting your efforts okay so now today your love is going to be definitely increased because we are going to introduce you one of the most powerful extension okay which is a machine learning and why I am saying this isn't powerful I am going to explain to this part so as now the picture blocks are interfaces here so I am going to add an extension and you can add the extension with this add extension button on the bottom left corner you can see this is add extension okay and once you are going to click it here all the extensions will be available okay so we will just do it again so you can see on the left bottom corner there is a button which says add extension so you have to just click on add extension okay and you will be able to see different extensions so today as we are going to work with machine learning so we have a dedicated machine learning extension which requires an active internet so I am just going to click on this and this is going to add some options here okay so some of you might be definitely astonished to see that okay there are no blocks here how we are going to make this right because generally what used to happen if we are adding certain extensions so we used to get some blocks but here if we have added in machine learning extensions we have got only two options but not the blocks and let me tell you this is the power actually so here we are going to create our own custom models so till yesterday we were working on the models which were already created right now this is the time that you can create your own models so how we are going to do it I will just request you to like just follow me also if you are not able to follow me we will be sharing you the video okay we will be sharing you each and every details so you can do it later also so now to start with so you can see here when I added a machine learning extension there are two options the first option is create a model and second is load a model so we will use it one by one so of course the first part is create a model so I am going to click it here but let me just bring your focus here that you can see still I don't have any blocks over the pallet right so I am going to click on create a model once I clicked on create a model it is going to direct me to the teachable machines so I will just ask with is my teachable machine screen is visible yes yes yes okay so now this is a Google platform some of you might have already worked some of you are very new but we are going to take care and we are going to do the things with very very basics so even yesterday there were some feedbacks that I just go very fast but today I will just take care we will go with a very medium pace so that everyone is able to understand the things okay so the first part here as you just saw that in pictor blocks I clicked on create a model okay and it directed me towards the teachable machine so let me just tell you one more thing here otherwise you will be struggling that that the educators or the people who are trying to do this on their smartphone they won't be able to create them okay but we have a option to load the model so we are going to help you to work with these complete setup as teachable machine don't allow you to create the model in a smartphone so right now you won't be able to create it but you will be able to use the model and we will be directing you that how you can do that others who are using it in or I should say who are using it on the desktop or the laptop they will be getting this page here there are two to three options so first is image object, audio object and pose object so we can work with image or project okay I'm sorry I just it is image project project and pose project right so here we can work with image project and also pose project today we are going to work with image project so I'm just going to click on image project now here as you can see there are two model systems first is standard image so we are going to choose this okay after clicking here you will be able to see this part of teachable machines and this is you can see some lab setup thing right but we are going to work in it with very ease okay I'm going to explain you each part here so starting with the first part which you can see is the class so you can see there is a class 1, class 2 and there is an option to add class so you can add classes as many you want but of course you should have enough sufficient RAM and all those things because if you are increasing the class that is going to increase the number of images also I will just delete right now and we are going to work with two classes so what is class so if I just take you to the thing which I think one thing you can do is you can zoom on the thing little bit on the browser yeah so yes now this is perfect now now you can zoom yeah so now you will be able the classes okay so now what exactly the class is if I will take you back towards what how exactly we learned right so we understood the first things with the different pictures which were labeled so class is nothing but a label for the set of data you are going to give so now as we know that we are going to make a mask detector right so in mask detector there can be two things whether it a person will be wearing a mask or whether the person is not wearing a mask right there is no only third thing so how we are going to make it I am just going to write the name as mask okay so here I will give the data related I will give here to give the data where the person is wearing the mask I will also give the name of second class with no mask okay so here basically that we are going to give the data or we are going to give the images where person will be not wearing the mask now there are two ways to give the data so you can either upload it using this upload option and you can upload it from your drive or from your devices otherwise there is an option to take it from the web camera also so I am just going to click on web camera and you will be able to see me here on the camera right so now you can see a button I will just scroll it up a little bit so now you will be able to see a button which is hold to record so what exactly it is and it is really some of you are going to love it because it is going to give you a lot many images just on one click okay so till that time if you are pressing that button it will be taking the pictures okay so as it is written as a mask so I will just put on my mask here okay so after putting my mask on I will just okay so after putting my I think my voice will be audible to you is it audible Pankaj? hello yes yes walking okay so like right now just hold the mask right now I am going to press hold to record and this it will start taking the images so what the thing I have to ensure is that I give my picture with different angles so that my machine gets a good understanding right so I will just start with hold to record okay so you can give a good number of images but I would suggest if you are giving images around 200 to 250 that is more than sufficient yes so let's stop it here also okay so now I have given 241 images okay now what is the second part so I will just remove my mask okay and I will just click on web camera again and just let me scroll up here so now I will just give an input where the person is looking if he or she is not wearing the mask so you can give the images in different angles so I have given the images here also 242 so now I have given 241 in the upper one and 242 in the second one so here also you have to take care one thing that you don't give images in imbalance manner that like if you have given 50 images in mask and if you are going to give 100 and 150 images in one so it will become in balance okay so there might be chance that it can't work on a perfect way so try to put images equal you can put okay so now this was the first part which we just saw that gathering the data okay it was like gathering a training data now we need to train it so how we can train it is you can see there is an option to train a model as soon as I am going to click it here it starts like preparing the training data also you have to be a little patient because the algorithms are working on behind they are calculating certain things so just to give you a little idea here that they are going to so finding something like this so you have to click on weight because it's like there might be insufficient ram or like the size of ram becomes high so you have to take care of it also if I tell you a little about how exactly the calculation is working so the machine is going to find the similarities between all the images in the mask all the similarities between all the images in no mask okay so there would be one calculation something like this and the second would be like it will be finding the I should say the difference between both the set of images so it will take a little bit of time till the time we can take like pool as well as we can ask you can ask us a question if you have any I would also tell some parts here as well so the classes that we are defining here is something which is like after the training the model can actually take the image and then it will tell you whether the image is of mask or no mask so it is like a classifier a classifier based machine learning model which classifies it into an image into different classes so what you can also like do here is someone was asking about what if we are not wearing the mask properly so what you can do is you can just add some classes into it like someone wearing mask till here or someone wearing mask here on the so those kind of like in machine learning there is one rule which is very important the more data you give the more better the machine would learn okay so it is like us only if we have more experience then we are able to do the things better so similarly in machine as well if you want to give more data then you can like you will get better result as well I will just ask you to just do one thing meanwhile this is happening you have shown the option of uploading the image from web cam there is also an option to upload the image as well you can just click on it and just show them how they can do it not doing it fully but showing them so I will just show you here like if you are going to click in here on upload okay so it is you are going to see something like these two options okay so in these two options it is the first is choose the image from your file okay and the second is import the image from the google directory if you are going to click it here it is going to open the library of your device and here where exactly you have kept on the images you can just select that image and you can directly put up also if you definitely there are there would be some 100 or 150 or 200 images so just select all even if you are going to select all the images at a time okay and you are going to click it open all the images will be coming to this part okay so I will just show you this part again like give you a clear idea if you are going to click on this place it will open up your pc library or I should say your hardware where you have to just select all the images okay like I will just show you like if in case it might have been something else okay so I will just show you here I will just go to desktop like you can import like if you are making an animal classifier then you can import the images of cats and images of dogs and then it classifies accordingly and then create machine learning models for identifying different between a cat and a dog so this is something which you can do very easily just select all the images it will import it and then you can just make the training model as well yeah I think Ayush you can just show them the second part one question yeah I will just take one question which was is it mandatory to that all the images are in same dress so if you want to create versatility in this then please have more different variety of images like you can try different colors of mask as well so that your machine learning model is more better at identifying those particular yeah yes Ayush so yeah so now this was the second part right which we just did the first was like gathering second was the training so now the third was about the testing so now we are testing it and now you can see it is showing me the answer is no no mask if I am going to put my mask here you can see it is showing mask right and as soon as I just remove it you can see it is showing no mask if in case you are not satisfied with the testing there is no problem just add more data start training again and just check it again okay but as I said like you can do it very easily just in one go if you are going to give somewhat like one I should say 200 to 240 230 images now these were the first three steps now the fourth step which is a very major step comes here is of deploying it means that we have created a model but how we can use it right so we can do it very easily we have to just click on export model okay so you can see on the top of a testing block okay there is written as a preview so here it is a export model option right so when you are going to click on these export model it will open a new pop up and here you can see like the this is a description of the model which you have created so here we are going to upload the upload the my basically I am just uploading my all the calculations to the teachable machine because definitely I don't want to take this heavy things all the 241 to 242 images things in the picture block so what I am doing is I am just putting those data not the data the calculation into the teachable machine and I am asking the teachable machine to get me a shareable link okay so this shareable link we will be using okay so you can see it has mentioned like your shareable link and it will take a little bit of time like you can totally understand it's a calculation of 241 and 242 images also like you might have seen what is one thing that there was there was 50 iterations so 50 times this was okay so now we can see the link has arrived here you can see it has mentioned as your shareable link and here you can see the link so I just need to copy this okay I just copy so now it is coming as copied and I will go back to the picture blocks okay now just again ask like is my screen of picture blocks visible yes yes it's visible okay wonderful so now the part comes to deploy so as I said the users who are using it in the smartphones won't be able to develop it but will be able to deploy it right so we are going to share you this model link and how you are going to use it now the things will get same for the people using in smartphones and the people using in desktop so I will just request people to put a model link on the chat box so that the other like other users who want to do it in their smartphones can also do it now we have done the first part the second part is we are going to click on load model so here you see there is an option of a loader model so then you are in maybe you can just tell them how to add the machine learning extension once again so you can add a machine learning extension just by using this add extension button on the bottom left corner so you can see if I am hovering it over it is coming as an add extension right so if you are going to click on this add extension it will be coming up as a machine learning whenever you are going to click it these two options will come up here right okay so if we are just half way completed if you have any questions till now please ask to us so that we can help you to understand the part still here then we will be moving to the deployment part so Pankaj if any other questions are also we can take it I think so the one question was like how to add the extension that is already answered the next question was how to get the shareable link so maybe you can just show how to like the export part once again because that would be important and one more thing if you are a mobile user you won't be able to do the teachable machine part in your mobile but what you can do is you can like use the link that we are providing of the model in the teachable in the pictor blocks to get the teachable machine model so that is something which you can definitely do and if the model is training currently so you have to wait for some time for the training to get over or you can reduce the number of times you are training you can train it for 10 times also it's not that you have to train for 3 times yes you can just show them once again yes so is my teachable machine screen is visible is my teachable machine screen is visible yes so I will just show them the part again okay so this was a testing block in the teachable machine so we clicked on this button which was export model okay when I clicked on export model it just brought up something here okay so I have clicked on this this block basically which says update my cloud model so it was first upload my model upload my model to cloud so we have already done it so right now it is saying update and after that there is a place here which is saying your shareable link so you have to just copy this link okay just by choosing this option of copy and then you have to move back to the I should say to the pictor blocks so do we have do we have any other like questions because this part as I said like once you will be able to create this it will become very much easy for you all if you have any questions please feel free to ask yeah 95% of people have their doubts clear so maybe you can move ahead and people who are missed should move the thing can go back and just go back to the video and just see the thing and get their doubts cleared yes so now as I said the part of like deployment or I should say using that model using the link will remain same whether you are using it using it in your smartphone or you are using it in your desktop so what you have to do is you have to click the second option okay the second option is load our model so you can see an option of load our model okay so you have to click it here on load our model are you yeah is my screen of pictor blocks visible are you sure the internet is little bit unstable just give us a your internet is little bit unstable okay I will just check just give me a moment yeah now you are very well visible you can proceed now okay so I will just reshare in a moment okay I will just reshare okay so I hope my pictor block screen is visible now to everyone yes okay wonderful so now as we were into the part where we were deploying that model using the link right so we have to click on load our model option once you are going to click on load our model option you will see a popup which says load the teachable machine model and here you have to be very precise sometimes what happen you think you have pasted the model because on the back it seems like htps teachable machine with google model okay but you have to paste it here and it will become little darker and you will be able to see this just moving of a cursor okay so you can paste just by clicking on paste option now after that you have to choose the model type so currently as we trained our model using the images so I will keep it chosen like an image now I am going to load a model so now the magic is going to begin basically so now it's loading also I would like to bring one query for like you might feel that sometimes you are going to see it is saying loading fail to load the model so if you are able to see this that fail to load the model just check that you have given the link correctly or not and now here on the screen you can see your own customized model now with the pictoblocks block system right so now here we are going to use this block and going to create a script for a mask detector so the first step so the first step is going to we are going to use the block which says when green flag is clicked okay from the events and now like if you want to just think of that how do we want to how do we want it to happen of course if you want if we have made this mask detector what we want here is that it should be able to detect the mask in real time okay so like we have already worked with this so I am going to I just first zoom my blocks a little bit I think now it will be visible much more clearly okay so now yeah so I think this is clear vision yeah so now what am I going to do is I am going to use this block so I just yeah so I am going to choose this block basically you can see it is the same which we have used previously also that is turn on the video on the stage are you may be you can are you yeah yeah yeah am I audible Pankaj are you maybe you can turn yes you can turn off your camera I think because of your camera you are offering my internet okay sure should I just do it in a while okay so is my screen of pictoblock visible yes yes okay so now as we were talking about the first block that was turn on the video on the stage with 0% transferency so now you can see I have came up on the stage okay and you can see me hi everyone okay so this block was basically for bringing my video onto the stage okay so now the next part is we have to find that whether I am wearing a mask or not right how can we do that so exactly like if we are talk about that we want to implement this thing in somewhere on the roadside or somewhere in the market so there can be chance when the person is removing the mask because I need to tell them you have removed the mask right so it should work on a real time and that's why I am going to use a forever so that everything goes in like a real time and now I will create two conditions okay so for creating it I'll just go to controls and I'll just bring this if so now the first part so you have to listen to me very carefully it's like if I am wearing a mask my machine should say thank you for wearing the mask right the first step is if I am wearing a mask so we have an option in machine learning which is directly going to check this okay so I'll just pull the blocks above so that everything is visible to you all okay so now here you can see it is a block which is going to verify directly the class okay so I'll just choose it from stage and we will make leave it to the mask so let's just click it here so now you can see it is analyzing and let's see what exactly it gives an output so as like we have already discussed that every hexagonal block in the pictoblocks is going to give you a Boolean value that is it can be whether true or false right now you can see it just brought up that false so what I am going to do is I am going to put it here and if this above condition is true that if person is wearing mask I will go to looks so you can see the looks is the second one okay and I will ask my I will ask my machine to say I will just ask my machine to say thank you okay so thank you for wearing the mask right and I will put it here now I have to make it again the same way for I should say if I am not wearing the mask I will just put it here okay and I will bring the same block okay you don't need to worry other like you don't have to search different blocks but you will be able to see the same block I will choose stage here because I am currently on stage and this time from the drop down so you can see there is a drop down here okay in this block which I have just brought it so you can see it has a drop down so this drop down I am going to select no mask if my machine is going yeah so so if there is one issue that you might face while adding your camera onto the stage so that issue would be like your camera might be used in the other application so here the other application would be like teachable machine so you have to turn off the camera from teachable machine and then just try again on the pictoblocks and you would be able to see that your camera is working very fine on pictoblocks if you are facing any more difficulty on the camera just restart pictoblocks and you would be able to do it so I will just give you a short thing that what Mr. Pankaj just told you right so if in case you are finding a trouble with the camera so you have to just switch off the camera on the teachable machines because if your camera will be utilized there you won't be able to see it here right once you have closed it and then you can come to setting options okay in settings you have an audio video setting and you can hear from here basically you can select your devices and check the whether everything is working or not after that so I will just turn on my video back again yeah so we were on to the part if the problem is like if I am not wearing the mask again I am going to go to the look so I will just stop this so I will just go back to the looks and here this time I will write please I am sorry please wear your mask so once I have given this I will just put it here and I will put everything inside the forever so forgive you a little recap okay if you have not understood so basically what we have done here we have just used this block the first one which is turning on the video on the stage then to check the things in the real time I have used the forever block now to check whether I am wearing mask or not I made two conditions with the if and if it is found that I am wearing mask my machine is going to say thanks for wearing the mask if in case it is found I am not wearing a mask so you have like your machine is going to put it something like please wear your mask so now when you will be making it so you can remix it with your own creativity so you can even do something like please wear your mask otherwise you will be fined for it right if you are not wearing a mask you won't be allowed to move into the hotels or restaurants so in this way you will be able to do it so I just add all the things into the forever and let's see whether it's work or not so it is asking for please wear your mask but as you can see it's not work so the second thing very important is you should have a proper light you should have a clear background so I think so you can see it saying please wear your mask and if I am going to wear the mask it is going to say thanks for wearing it so it also depends on somewhat like the camera thing but if you are going to put a good amount of light I will just try to do that okay so then in that case the mode production would be there you can instead of detecting it from stage you can detect it from the camera that would work better okay so I will just do the same I will just stop it on your stage the other tovi is also coming so that is something which might also have problems as well so now okay so now it's working better okay yeah so here you can say this is a machine learning okay and you can see here that it is saying no mask and if I am going to come up with the mask it is saying the mask 99% right so now you can see that how exactly the well it is working okay so we would like just I will stop this here if you have any doubt please feel free to ask I will just stop this here okay so please feel free to ask and I will just keep this thing open for you so that you can just see around the script what it is if you are feeling any doubt you can ask to us and also we are going to just show you some of the wonderful implementations of this right so it's not only that we are just working just on the like computers and I should say on the phones but you can increase or I should say you can also bring the things into the education where these things will become much more highly creative and understandable so we are going to see those things okay so if anyone have any question we have the question so one question is how to bring mask and no mask block so I think the answer to that is you have to import or the load the model in picture blocks in order to get it so just click on the load model button and just import the model with the link that we have given to you so you can find the link if you have created the model yourself you can copy the link from there otherwise you can copy the link from load the model as well like from the chat that we have given to you so that is something which you can do and the second part was how to get the recognition window so you can get the recognition window when you just select the tab webcam it will get the recognition window automatically so the recognition window actually identify like take the web feed and identify the class accordingly as well so this is something like it is identifying in real time and if I use shift you wear a mask just to showcase so in the recognition window you can see the percentage of the confidence as well okay so this is something that you can do so your teachable machine link that you create will be valid for a very like till till right now you can use it like it not being deleted on a regular basis so you can use but there is no confirmation from them for how long it would be there so that is one part and one more question was like how to save the program someone was asking so Ayush maybe just you can say tell them about that so as like we have discussed this before and I will just show you again here like there are two very major good features so you can save it just by using save as option okay so you can just click on save as option you have to give the name and the location so I hope my teachable machine like hope my pictoblock screen is visible so here you can just find the location where do you want it to be saved and the name okay and you are going to click it save I will just save it to show you even the notification so let's make it ML model AI okay so if I am going to just click on save so it is going to get saved and you can see the project saved and even the ML the name up name come up here also if you want to generate the link so you can just click on this and this is a wonderful feature where if you are going to just click it we are going to get the link of a complete file okay and you can share the just share your complete file just using the link so I have just like I wanted to say this to others also like we have been seeing that even in some assignment you have shared as this file so we want you to share this file okay so this is just for file sharing to others like if you have made something if you want to share it to the students okay so you can do it very easily yeah yeah so there are other questions as well so one question is if you have multiple camera system connected in your picture blocks then you can also change your cameras as well so if you can just show them how they can change the camera so you can change your camera with going into the settings okay and there isn't fifth option which says fourth option sorry the audio and the video settings if you are going to click it here this pop up will come come up here okay so on the camera you have to just click this and you will be able to see the multiple cameras on device and you can select from the list okay so you will be able to select it from the list so yeah yeah do we have any other questions yeah so there are some other questions as well so like why is the like mass recognition window telling only 77 or 77 or 98 percent so not 100 percent so when we talk about machine learning so in the machine learning there is a concept of confidence okay so the confidence is something which the computer tells us on the classification that is doing so the computer the confidence is something which computer gives and tell us that okay if you are working in a situation like identifying an image and classifying the image I am this much person confident that okay this will happen this is this class okay so getting 100 percent confidence is very possible when you have like very good artificial intelligence but it is not ideal to have 100 percent for a class it should be that like if you are getting a result above greater than 70-80 percent then it will be working very like your model would be working very fine as well so it is important that your model is being trained in a very well manner the classes that you are giving are appropriate and the data as well so that you are able to get the model like trained in a proper manner as well I have two cameras I have to disable one in this side no you don't have to disable the camera on the device manager you can just choose the camera from the setting audio and video setting and then you would be able to get the data like camera as well so you can just select it from this drop down and you will be able to switch the camera in between machine learning this is provided are the latest images from the camera also make use to bring them so there are some processes in which you can use the current images as well so the current model that we are using is based on supervised learning so where we are giving the machine supervised data so like giving data and the level as well so in this particular type of learning the training part is done separately and then the testing part and the deployment part is done separately so if you have to add more data then you have to add more data from the other in the training section not in the testing section so those are the different type of machine learning which we will not cover during the session but you can explore them as well if you want to turn off your camera just click on turn off the video on stage so on the block you can just select the option of off and then you can turn off the camera as well and if you just click on the turn on the camera it will turn on the camera so someone has asked that your machine learning is saying in class 2 so if you are like if you are not getting mask and no mask option then most probably you have not renamed your class on a teachable machine so you can go back into teachable machine and rename your class as well so it can be done by going back and then you can click on mask and no mask in order to rename it maybe Ayush you can just show them about the other part and take out the Q&A so we will now just move to the I am sorry so now we have done this part where we just saw how we made the mask detector now if we talk about the applications of machine learning it has a very wonderful applications from image classification so one of them we just did from fraud detection post detection in health care you can see in gaming in using cellulances, recognitions and also the recommendation system so these are some of the wonderful applications of machine learning and now here comes the part which even I was waiting so I would just ask Mr. Pankaj to show that how we can make this machine learning more interesting with some of the implementations yes so I will just show you my screen again I hope you are able to see the screen so before going further I will just take out one more question that is coming like you have trained the model yourself on your face and if someone else is coming there and do the thing then how will it perform so we have to understand that the machine learning model if it is trained on some particular type of data and if you are performing the action on the other type of images then you have to have the machine might not work in a very proper manner you have to it won't be very reliable so for that what you have to do is you have to gather a lot of data from other people and then train the model so that model won't be more confident it would remove the biases and it would remove all the things from the model and then it would give you a better result at the time of deployment so what you can do is you can have 4 or 5 people give their data or face it with varying mask on different angles and then without mask and when you train that particular model that would work very better in a versatile situation and if you have to do it that will give you better result than training the model in a one situation on you so this is also very important when you are learning about like the stage table machines so as you can see so I will just take a bit actually some are asking like the screen is not visible so I will just request educators can you please respond on the chat so is the screen visible because we definitely don't want you to miss the screen I think this screen is visible very much visible coming to the next part here you can see some of the projects that you can create like cat vs dog classifier and even if you can make things like classifying and getting pneumonia out of it so pneumonia from the patient actually so this is something which is very exciting as well and you can try to make some projects as well the next part that I wanted to show you is on the hardware implementation of it so this is the project done on quarky so as you can see here we have the quarky robot and we are training the quarky robot on identifying whether it is seeing e-waste or like paper so as you can see here there are three classes that we have created it shows paper e-waste and background when the e-waste comes battery comes it shows the e-waste and then when the paper comes the paper is coming so once we have done the training part what you can also do is you can implement the machine learning models in the hardware world as well so what you can do is you can create some problem statements like you can create your own situation where the robot which has to identify the type of waste and go to a particular like a checkpoint so as you can see here you have the checkpoint 1 and 2 the robot starts from here it identifies the paper waste and it is moving towards the checkpoint 1 so it is doing line following robot line following as well it is doing machine learning as well and it is also doing the identification card so as you can see here the robot has stopped here because when the robot reaches to this particular point where the checkpoint is coming at that particular point it is actually analyzing the image again using another extension that we have in pictoblox to identify whether it is looking at number 1 or not it identifies these cards and it makes like it stops at that particular point as well so creating like we are also working towards creating real world applications out of it as well and whatever you have learned in the machine learning you can also implement it with the robots as well I would also want to want to show the video of the video as well so this video is of the quirky innovator kit that you will be getting on like few of the educators will be getting on completing all the assignments as well so I would just start the video it is an AI and robotics learning companion for the next generation Pocky takes children on an exciting journey to explore advanced concepts on AI and robotics with an easy engaging and hands on approach it is packed with interactive online courses and engaging live sessions that help children to learn, code and make their own real world projects in other words one quirky endless possibilities quirky is portable carry quirky around and control it using your laptop tablet or even a smartphone it's easy to program whether you're a coding newbie using blocks or an expert using python comes with a plug and play interface making it hassle free and with some of the coolest latest tech so imagine a self driving car which identifies sweet signs and pedestrians or a faithful robot dog that follows commands with the help of machine learning and object detection it's evident that quirky is for everyone who wants to be ahead of the curve ready to explore the universe of AI and robotics with me yeah so this was called quirky so quirky innovated is something which we are giving to educators who perform all the assignments and who give us like provide all the assignments as well I think there is some problem that we are getting on the screen share so we will be telling you more about quirky in the coming sessions so please don't be please wait for some time we will be telling you about more on quality in the coming session and how you can benefit from it in the robotics as well today we just wanted to show you what thing you will be getting when you are submitting all the assignments we want you to motivate as well so it's something which is very excited for us we have been working on it for a very long time and you can implement real world artificial intelligence and machine learning based models on it as well so it's a very exciting thing so yeah I wish you can go to the slide of the assignments and tell them about the assignments and everyone who have queries about quirky we will be taking out the queries about quirky in the last session and today we wanted to just show you a glimpse of what you will be getting when you are submitting all your assignments and so if you submit your assignment you will be choosing 7 educators from all the people who submitted the assignments and give quality to them so yeah yeah so thank you Pankaj just showing such a wonderful video and it just amazed every time I just see that video that how exactly the quirky is working so now also is my screen visible yes yes yes so also educators like if in case you want to explore more things okay if you want to explore pictoblogs in much more ways so we have a dedicated stem project hub in our website okay so I will just request local to put the links okay and you can come to come here and just explore the different projects related to artificial intelligence and machine learning and different hardware also okay so we don't do and this is a complete tutorial where you will be getting step by step instructions for everything also we would just request you to please just subscribe to our stem media channel because you are going to see wonderful project videos there and also we will be notifying you about the sessions and the upcoming plans what we have what we have planned for you okay so just please subscribe the youtube channel press the ring bell icon which will every time notify you okay and also explore it we have a different dedicated playlist according to your need okay and now coming to the assignment part so here basically as like you already know today we have made a mass detector so we want you to make a mass detector okay and mass detector record a video and then just upload that video to the youtube google drive and give us a proper access to view it just paste that link to the form which we will be sharing you in a minute okay so I would just request you to if you want you can just scan the QR code you can take down the links from the chat box also we will be sending these links to the telegram and your email IDs but make sure you are giving your assignments completely and trust me don't miss out the chance to win the educators batch the certificate and also win this robotics wonderful robotic kids and also still you are not a part yeah okay and if you are if you if you want to submit anything other than the mass detector you can do it that as well we want you to explore the machine learning thing so like it's just that you have to create a model which a machine learning model and then simply load it in pictoblocks and create projects out of it so mass detector is something which we which you can you should try definitely and if you are also want to implement something more out of it you can do it that as well yes Ayush yeah so all the users who are doing it in a smart phone don't forget to take this link with you you have to just load this and you can still you can put a creativity so we have talked about that how you can put a creativity so you can put something else in the like the save lock and the different things you can use right so now if you are not still a part of a telegram channel so please just yeah still you are not part of the telegram so join us by using this QR code also we are pasting the links on the chat box and in the next session we are going to learn about natural language processing and that is going to be very interesting also we would like to tell you that we have taken questions but you have so we will welcome you and now we would also like to tell you like if still in case you have any query related to anything if you have business related queries so you can mail us on the bd at the Stempedia if you are finding any problem in your projects or any support you require so please email us on support at the stempedia.com also you can mail us for the curriculums and the different activities if you are interested on the education at the stempedia.com now I would just request the educators if you have any questions we welcome you all to ask those questions so I will just stop my share here so yeah so how to turn off the camera in the picture block so Krishna you have to just click from the drop down turn off so you have to just click off from the I should say from the drop down of the first block and like if you are finding any problem I have mentioned I can't save so just share us the screenshot on the telegram so that we will be able to help you yeah any further issues you face just send us on the telegram and if there is any technical difficulties then just mail us on to the support at the stempedia we would be taking care of all your queries there as well yeah and we will be sending out the link of all the previous assignments to you on the emails so please look at all your emails you will be receiving the email within an hour and after that you can work on the assignments you can work on the previous assignments as well so once again I would like to repeat that when you complete all the four assignments so like three assignments we have already given up to you this is the past assignment and the third one is today's assignment so you have to submit all the assignments in order to get the batches of the educator batch and the certificate as well so it's very important that you do it if you have exhausted your like in this machine learning example like project you don't require any credits and if you are facing any problem with credits then contact us on support at the red dust and we would be taking your queries from there as well so any any problem you are facing just email to support at the red dust and we will take care of all your queries as well and the last date of submitting the project would be 2nd of July but we would request all of you to submit the project within 24 hours so that you can work on the next concept and the project so it's very important that you are very regular and get the project done as well in this assignment also you have to make a video of the working project and this will help us judge whether you have completed the project on your own or not so we are getting little strict here in the assignment part so if you have submitted the wrong link and the previous assignments then please resubmit the previous assignments as well with the new links so that there is no issue on getting the certificate so yeah we will be sending out the feedback about the last few assignments on the email so we will do it very like we will do it and we will notify you whether your assignment is approved or not yeah and we will be sending you the recordings as well so please wait for our emails when our emails will like in our emails we will add all the instructions about the all the things would be added into the email so please look into the email that we are sending and yeah I think we are already 5 minutes up from our time so we would like request all of the educators to put their queries on the telegram group and if you have any technical queries you can put the query on support www.centredia.com and once again thank you everyone for giving us time on machine learning it was very good to see the response coming from the educators and I am very excited to see when you implement these projects with the students and how they react I have personally done a lot of sessions with the students and they react in a very very impressive very fruitful manner so I think you will also get the same results when you are implementing it with the students thank you yeah so thank you again to all the educators who have joined us and hope you have enjoyed the session still if you find anything to be improved please let us know and also please don't forget to submit your assignments we will be sharing you the links of the submission forms in your emails as well as on the telegram group and be ready to start a new journey again we are going to do a new topic that will be the natural language processing so don't forget to join on the same time and yeah thanks again thank you everyone thank you