Object Detection and Characterization of Traffic Live Video Stream using Machine Learning

Video has become a mainstream media source on the web, and live video streaming is growing as a prominent player in the modern marketplace for both businesses and individuals. User?s view of watching a video is growing increasingly and in 2019 the rate is 14.5% and is expected to be which will be ro

Project Title

Object Detection and Characterization of Traffic Live Video Stream using Machine Learning

Project Area of Specialization

Artificial Intelligence

Project Summary

Video has become a mainstream media source on the web, and live video streaming is growing as a prominent player in the modern marketplace for both businesses and individuals. User’s view of watching a video is growing increasingly and in 2019 the rate is 14.5% and is expected to be which will be round about 16.2% in 2023 with video comes the term streaming.

Streaming is what happens when consumers watch TV or listen to podcasts on Internet-connected devices. With streaming, the media file being played on the client device is stored remotely, and is transmitted a few seconds at a time over the Internet.

The technology that proved the development of communication technology for human needs is a camera also known as CCTV (closed circuit television). This project will be used to get the roadside data of different types of objects using a camera that is attached with a Raspberry pi module. [Raspberry pi] a tool used as a web server and is installed by motion eye as a web interface that is integrated with a camera (CCTV) & send it to a server computer for processing and analyzing i.e. object detection, object classification, speed calculation, number-plate checking, road condition through Deep learning AI. The project is unlike of the conventional type that takes data from the camera and send it through wires and antennas, rather it is a live streaming over internet and the data is send using internet protocol(IP), where each unit has unique internet protocol (IP). The data in the process will be analyzed through different libraries used for video streaming and computer vision, named Open-CV and TensorFlow open source tools. Open-CV provide a common infrastructure for computer vision applications and to accelerate the use of machine perception in the commercial products and TensorFlow an AI library.

Project Objectives

1) To define IOT device, Pi camera in order to capture the live video streaming.

2) To process the data extracted from the IOT through intelligent Algorithm.

3) To extract the statistics from the data taken from the IOT device. 

4) To transmit the analytics taken from the data to the receiver end for making a better decisions for the community and environment.

Project Implementation Method

The data will be taken by the camera installed on the roadside, taking the live stream this data will be sent to a server using internet streaming using a unique internet protocol (IP). This protocol will be using the services of network over internet protocol (NO- IP), in order that the synchronization between the camera and the server is not lost. On the other hand, if the NO-IP is not used it will end up with a problem of changing the public IP at each restart or system fault, which will result in the loss of synchronization, the objects will be detected from the road side using Deep Learning AI(Artifical Intelligence) algorithms (YOLO-COCO) i.e using YOLO(You Look Only Once) algorithm for COCO dataset. COCO dataset contain 80 different items(Car, bike,tie etc.) that can be detected by the YOLO(You Look Only Once) algorithm in our project. Further there will be a sensor (optional) attached with the camera that will detect the emission of the vehicles and send it to the server. To the server there will be an attached Computer which will be pre-available with the library Open-CV. The data will be processed using the Deep Learning AI algorithms & FF-mpeg library taking out the useful information i.e. object detection, object classification, number plate checking, emission from the cars(optional), congestion and the road condition. The data processed will be passed on to the library Open-CV which will utilize the data in better way. Our system will also have the capabiliy to identify the car type, using spatial features collected through lidar(Light Detection and Ranging) sensor.

Benefits of the Project

1) In the previous version of the project the data used to be stored in the databases for analytics purpose. This project will be using live video streaming, so it will save the extra memory of the databases that was used in the past. The project will find the analytics at the run time.

2) Previously the conventional methods were used for the project, which costed more due to computation. Still achieving low quality results. This project will be using AI (Artificial Intelligence) technique YOLO(You Look Only Once), which is a very modern and a faster method to solve the same problem easily and more efficiently.

3) Previously the analytics from the data used to be taken out after the data was loaded to the databases. This project will take out the analytics at run time saving the memory of databases

4)This project will be used to take better decision for the community and environment.

Technical Details of Final Deliverable

The final deliverable will include:

1)  A hardware system containing a jetson nano computer with camera module that will be implemented on the roadside.

2) Documentation regarding the project (Implementation, precautions & maintainance guide).

3) A complete software behind the project that will be handling all the corresponding calculations and queries.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

IT

Other Industries

Transportation

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT)

Sustainable Development Goals

Sustainable Cities and Communities

Required Resources

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