 Good day everyone. I am Assistant Professor Rushad Mastri from the Mechanical Engineering Department at Valgyan Institute of Technology Sharapur. And in this video, I'll be talking about machine vision applications. This is a two-part series in which we talk about machine applications, machine vision applications in different aspects of industry, defense and so on and so forth. So the learning outcomes of this session are that the student will be able to list applications of machine vision and then explain applications in industry. So let us recall the definition of robot vision and robot vision, computer vision and machine vision. In fact, we had done this in the previous video. Like I said, in many cases, I've seen these definitions being used interchangeably. But I would define computer vision as the study and application of methods which allow computers to understand image content and machine vision is the application of computer vision to industry and manufacturing. And robot vision in very specifically is the application of machine vision for industrial robotic applications and now for mobile robot applications. So here I've taken a very simple application of machine vision and this involves pipe counting. So this, in fact, from the website of a machine vision software developer called Robo Realm, they are third-party developers of image processing and machine vision software. An important aspect when it comes to machine vision application is when it comes to quality control and inventory management. So this is one application in which machine system has been used to count the number of pipes in this particular setup. So very simple tasks. The machine vision, what this particular application does is, in fact, it uses what is called as edge detection in order to find out the edge of the pipe. Then it sharpens the edges, refines the whole process, draws a circle around the pipe and then uses an algorithm to count the number of pipes. So it's quite a straightforward application implemented in a very simple, at the very robust way. So these are the kind of applications you would come across in industry very frequently. Another application from the same source is on object classification. So what this particular example tells you is, what the software has done over here is, there are three types of poles, in fact, that you can see on the screen. This is a red bolt over here. These ones with the violet color, these are for now a different size. And then you have these green colored bolts. And what the software has done is, it has automatically categorized these bolts separately, and it has also counted them. So it has counted one red bolt, seven green ones, and I think three blue colored ones. In fact, it looked purple to me. It's blue. So this is, again, a very typical application of machine vision in automation and industry, where they are used for object identification and classification. Another important application in quality control, and this is in food processing in industry quite frequently. So in fact, if you watch National Geographic for in mega factories and stuff where they show how it is made, you will see applications of machine vision to do something like this. So in this case, it's actually what the machine vision is trying to do is find out if this coke bottle is filled up to the correct mark. So it's based on a very simple edge extraction screen. And if they see that is up to this particular level, a simple metric is used and simple arithmetic is used. And it's basically a very simple pass and field task. So again, a very classic application of machine vision that you'll find in industry. Now, when it comes to semiconductor industry, it's one of the most, you can say, extensive users when it comes to machine vision applications. So here, you'll find applications which deal with inspection of silicon wafers for microchips, for the integrity of components such as resistors, capacitors, and in fact, even regarding their connections, the leads, the solders, if they are correct or not, quality checks regarding PCBs and so on, a very extensive use of image processing done in the semiconductor and electronics industry. To examples here, in fact, it's basically for a dimension check for an electric socket. And the other one shows an image processing trying to ascertain whether the PCB has been built correctly or not. So like I said, this is faulty component identification electronics is the same image. In fact, is a very widely used application in the electronics industry. Machine vision is used to identify faults in components. Sometimes even during operation, high amount of thermal energy is generated and infrared cameras can be used to capture these faults. And infrared imaging is also, but this is very specifically using even testing and even during actual usage as well as a part of diagnostics. A very, very popular application of machine vision is part identification and categorization based on 2D and 1D barcode reading. So which we call QR code nowadays, which is quite popular now for payment gateways and stuff. So verifying moving parts to fabrication process, they typically use a QR code and vision systems can do this at several tens of pieces per second. Then product identification is sorted. This is again done with the help of barcodes and this is where machine vision systems come into picture. They also use the same for date verification, lot verification. And if there's a problem, then that can automatically flagged off. So again, this is a very popular application for machine vision systems. Like I said, this is again, another example of barcode recognition. In fact, barcode recognitions have become now so popular and so common, you find this in pretty much every supermarket that you find. In fact, barcode recognition and QR code definition is available in even in cell phones. It's now a very basic feature which is built into pretty much every mobile device. Machine visions are a bit different in the sense they have to be in a more of a rugged environment. So they are built differently and they are more rugged. But in terms of algorithms and functions, they are pretty much the same when it comes to any barcode recognition. In fact, the one at the right is it includes an optical character recognition as well, which is checking for dates and labels. So this is also one very popular feature nowadays for machine vision applications. Another important application is measurement that is gauging and verification of tolerances for floor detection and stuff. So in metal industries, it's essential to detect flaws and surfaces. It's very difficult for an operator to do the same. Machine visions can do it very precisely and very fast. Important application is gauging and verifying tolerances. So a lot of precise machine tools when they are built, a highly specialized and dedicated measuring system is used to check whether they have met those manufacturing tolerances or not. They prevent your maintenance. Can you think of certain applications in science and medicine? In fact, if you have visited any CAT scan center, typically that's a machine, sort of a machine vision application as well. So when it comes to scientific applications, two things crop up. One is in biomedical and bioscience applications, the other is physics. Spectroscopy is again a very popular application for machine vision systems. So spectroscopy measures light or electromagnetic radiation. And it's basically used in astronomy and remote sensing to determine chemical composition and physical properties of astronomical objects. In fact, there's a dedicated branch within machine vision, which is meant for astronomy and astrophysics. Similarly, another important application is in fact, microscopy. It's a topic on to itself. So electron microscopes are in fact one of very specific applications of machine vision systems. Another application in fact is biotechnology. For example, dental x-ray imaging, CT scans, scans of organs. So in fact, medical applications for machine vision systems are tremendous and they are ever increasing. This is again an example of what imaging when it comes to medical applications. So you have dental imaging, disease identification. In fact, image processing for medical image is again is a very specialized field. Okay, then you have very dedicated algorithms, which which can extract information from even poorly developed images. So it's in fact, it's a very well developed field of motion vision, which comes into medicine. Another important application is like, for example, detection of tumors, etc. This is an example of a brain tumor which has been captured by a camera and then image processing. In fact, artificial intelligence now can detect tumors with more than 95% accuracy. In fact, it exceeds that of top top radiologists and technicians. So mammograms typically they are analyzed using image processing such as segmentation shape analysis, enhancement feature extraction and then matching. So like I said, the best resource for this are online resources. And there is plenty of information available on the internet from manufacturers themselves. And I would definitely encourage you to look at the website of these manufacturers for very relevant and very detailed information on regarding applications of machine vision systems. On the other hand, the very classic book on robot vision is Four and Gonzalez. Some of the very basic concepts are discussed in a very simple and neat manner in this particular book. Other than that, if you're looking at machine vision and detail, there are dedicated books available on the same subject. With this, we'll call it a day for this particular session and we'll discuss some more applications regarding defense and surveillance in the next part. Thank you.