TRAFFIC CAR DETECTION LEFT-RIGHT

ABSTRACT:-

            Vehicle detection in Intelligent Transportation Systems (ITS) is a key issue making sure road safety, as it’s far necessary for the tracking of car flow, unlawful car kind detection, incident detection, and car velocity estimation. Despite the growing reputation in research, it stays a difficult trouble that need to be solved. Hardware-primarily based totally answers consisting of radars and LIDAR are been proposed however are too high-priced to be maintained and bring little precious statistics to human operators at site visitors tracking systems. Software primarily based totally answers the use of conventional algorithms consisting of Histogram of Gradients (HOG) and Gaussia Mixed Model (GMM) are computationally gradual and now no longer appropriate for real-time site visitors detection. )erefore, the paper will review and examine different car detection methods. In addition, a way of utilising Convolutional Neural Network (CNN) is used for the detection of automobiles from roadway digital digicam outputs to use video processing strategies and extract the preferred statistics. Specifically, the paper applied the YOLOv5s structure coupled with k-way set of rules to carry out anchor box optimization under distinct illumination levels. Results from the simulated and evaluated set of rules confirmed that the proposed version became capable of attain a mAP of 97.eight withinside the daylight hours dataset and 95.1 withinside the midnight dataset.

MODULES:-

• Image Capture Module:
  Image capture is done by video capture devices like cameras and then saved in convenient formats like mp4, mkv, etc. OpenCV is the library used to access the video  captured by the camera. Cv2 OpenCV library  is used to capture the video and pass it frame by frame for further processing.
• Vehicle Detection Module:
  TensorFlow is the framework to create a deep learning network that solves object detection problems. There are models in your framework that are available through the reference as Model Zoo. This includes a collection of pre-trained models trained on the COCO dataset, the KITTI dataset, and the Open Images dataset. Here we use the COCO data set.
• Vehicle Speed Prediction:
  The prediction of the speed of the vehicle was developed with OpenCV by manipulating and calculating the pixels of the image. The object is being tracked. On each frame, draw a rectangular bounding box around it. The speed of the vehicle is determined by calculating in pixel form the distance that the object has moved in a sequence of frames in relation to the frame rate and the total video time  is recorded.
• Object Counting:
  To count the number of vehicles, we used the number of TensorFlow objects. We use “cumulative counting mode” to count the  number of vehicles. In Cumulative Counting Mode, the vehicle speed  is calculated only  when the object crosses the ROI (Region of Interest) line, regardless of whether the system detects the vehicle.
  
  

HARDWARE AND SOFTWARE REQUIREMENTS:-

HARDWARE:-

• Processor: Intel Core i3 or more.
• RAM: 4GB or more.
• Hard disk: 250 GB or more.

SOFTWARE:

• Windows Operating System.
• Java
• R (3.4.1)
• R Studio