Page 145 - Trackpad_V5_Book 8
P. 145

meaningful information from those images, videos, and other visual inputs. Computer vision is
            important and linked to AI as AI enabled systems must interpret what it sees just like human

            vision and act accordingly.
            The goal of computer vision is to train machines to see, process, and provide useful results

            based on its observations within a very short time. This can only be possible if lots of data is
            provided to it which can be analysed over and over until it detects distinctions and ultimately
            recognises images. For example, to train a machine to recognise tyres, it needs to study a
            lot of images of tyres. and items related to it to learn to differentiate and recognise a tyre

            without fail.


                                                   Human Vision System




                                                                                   bowl, oranges,
                                                                                   bananas, lemons,
                                                                                     peaches




                                              Eye                       Brain
                                    (sensing device responsible for capturing   interpreting device responsible for
                                        images of the environment)  understanding the image content

                                                  Computer Vision System


                                                                                          bowl, oranges,
                                                                                         bananas, lemons,
                                                                                            peaches



                                 Input          Sensing device    Interpreting device       Output

            How does computer vision work?

            The two technologies required to accomplish this are:
               Machine Learning (Deep Learning)

               Convolutional Neural Network (CNN)

            Machine learning enables a computer to understand the context of visual data by using algorithmic
            models. To achieve this, enough data needs to be fed through the model so that the computer can
            teach itself to recognise images.

            Convolutional Neural Network (CNN) helps a machine learning model to process the image by
            breaking it down into pixels. It then transforms it into digital data by applying algorithms before
            comparing the captured images with those stored in the database. These systems are used to
            identify an individual based on their facial features like spacing of eyes, ears, chin, etc.





                                                                            Artificial Intelligence and its Domains  143
   140   141   142   143   144   145   146   147   148   149   150