Object Detection Flow:

     

So, in today’s blog I will explain the process of Object Detection, The first step for program to run is by user interaction to the system without user there is no use. We have developed chat bot already, from which user communicates with the machine, in this user can either types the name of the object they are looking for or they can upload an image to the search engine. We load another image from the database for matching the tagged objects from database image to load these images we are using OpenCV library function ‘imread’,  the first step we take is cropping or slicing. The purpose of slicing image is when we further check the images, we have very less area for program to run, I am making the slicing part as a function of our AI library, we will develop another function which is image smoothing where we partially implement OpenCV methods. After all the steps, We ‘Check image’ using OpenCV, if the image matches the database image we proceed to next step, where we use our matching Algorithm to check if there is similar looking object/s in the user image or not, while checking the image if the image doesn’t match the database image, then it will deleted.

Now here comes un interesting part of the process, Let say that after checking images, it fails ‘our matching algorithm’ then it does not means that the system has failed but have discovered something, the result we see is some new object that is detected which is similar to the tagged object in the database image. After all the procedure, we send this as API to the database, whenever user calls for objects through search engine we can send the tagged images to the ‘result’ back as a search result. 

Later using Machine learning we will make the process only limited to Few essential steps to the images processing and checking image will enhanced with our machine learning algorithm, the following chart displays the system's future process: