Computer Vision Based Real Time Detection of 2D Flat Surface Area

MAIN AGENDA Nowadays, the discovery of real-time objects and the size of objects (dimension) is an important issue from many places in the industrial sector. This is an important topic for computer vision problems. This presents an improved way to get things don

Project Title

Computer Vision Based Real Time Detection of 2D Flat Surface Area

Project Area of Specialization

Artificial Intelligence

Project Summary

MAIN AGENDA

Nowadays, the discovery of real-time objects and the size of objects (dimension) is an important issue from many places in the industrial sector. This is an important topic for computer vision problems. This presents an improved way to get things done once to computerize their ratings in real time from video streaming.

As we know that human eye is not capable of dimensioning an object as a computer with a peripheral vision can. The number of businesses adopting artificial intelligence grew by 270 percent in four years. The global artificial intelligence market is expected to reach $641.3 billion by 2028. 91 percent of leading businesses have ongoing investments in artificial intelligence. It is a high time to address the need in our local/pakistani industry as well.

WHY  AI - COMPUTER VISION?

Computer vision systems can help count stock, maintain inventory status in warehouses, and automate and alert managers if any material required for manufacturing is below demand or maintain an efficient yet accurate production. The computer vision systems can avoid human errors in counting stock. In massive warehouses, locating stock is difficult and much more.

Project Objectives

The main objectives of the project are:

• DATA INPUT:

The data will be in the form of video streaming in real time using a camera attached to microcontroller at a perticular distance of its focal length which captures the feed in stable conditions having a good view (with the help of light source) then will be sent to pre-process the feed.

• PRE-PROCESSING:

The feed in RGB color mode will the be converted into grey scale which avoiding the color noises as the computer vision algorithms works best in grey scale frame by frame.

• CONTOURING:

The data will then be contoured as contour is a boundary around something that has well defined edges, which means that the machine is able to calculate difference in gradient (significant difference in magnitude of pixel value), try to see if the same difference continues and forms a recognisable shape and draw a boundary around it.

• DIMENSIONING & SORTING:

With this data now we will be able to find the height and length as overall area of the object (Two dimensional) and here we need to adjust the camera frame as per pixel-per-metric whatever the metric be it in mm, inches or cm.

The final action is to sort the pieces which are not lying under as per production requirements and reject or approve those products.

Project Implementation Method

The main aim is to find a solution for measuring the dimensions of an object present in a picture using algorithms related to computer vision.To make this possible, we have to consider two types of pictures – first type is with the presence of object and the second type without the presence of object in the picture. One more important factor is that to know the details about height of the camera. We require prior idea on various libraries and frameworks, especifically OpenCV. In Linux OS or windows, Selected tool is available and it is an open-source software. In OpenCV, the algorithms for identification of real time objects, image conversion with various colour models to grayscale then getting it contoured to noise free the live feed and drawing a line to defind the dimension, all of this will require multiple camera window setup when implenting the project.

In this project work, even the 3-Dimensional object will be calculated and indicated by a 2-Dimensional area where the image background isinvisible. The 2D object representation being concave should not affect the output dimensions.

Benefits of the Project

   Object detection is a computer vision technique that allows us to identify and locate objects in an image or video. With this kind of identification and localization, object detection can be used to count objects in a scene and determine and track their precise locations, all while accurately labeling them.

   The manufacturing industry often struggles to get 100% accuracy in detect defects in their manufactured products, as it demands systems to monitor defects on a micro-scale (like monitoring the wrong threading). Detecting these defects at the end of the production process or after the delivery to the client can result in increased production costs and leads to customer dissatisfaction. These losses are comparatively far higher than the cost of adopting an AI- powered computer vision defect detection system in measuring object size.

   A computer vision-powered application gathers real-time data from cameras and using machine learning algorithms, analyzes the data streams and based on the predefined quality standards, detects the defects, and provides the percentage of deviation. Based on this data, setbacks in the production line process can be traced. This way production process can be error-free and effective. 

As it is rightly said that;

   ”The cost of not detecting a defect is much higher than the cost of detecting the defect.” Investing in a computer vision-based defect detection system can be a cost-effective solution.

Technical Details of Final Deliverable

A prototype of low-cost but more efficient solution to industrial productive defective system, consists of the following deliverables:

1. Raspberry Pi 3 range boasting a 64-bit quad core ARM Cortex.
2. The V2 Pi high resolution Camera Module with flexible cable.
3. A functioning prototype conveyor system with feeder.
4. An object detecting adjustable proximity sensor and LED light source.
5. DC conveyor/gear motor, and motor driver.

Firstly, the conveyor will send objects from feeder which be detected by proximity sensor to stop and send the image feed captured by camera to Raspberry Pi microprocessor-based controller for processing the whole processing from image pre-processing to contour to dimension object.

Final Deliverable of the Project

Hardware System

Core Industry

Manufacturing

Other Industries

Food , Others

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com