TRAFFIC AND ACCIDENT PREDICTION

ABSTRACT:-

                        Every year, about 1.35 million people are isolated from multiple shocks in  a traffic accident. According to statistics, between 20 and 50 million people are injured as a result. people lose theirs life through this type of traffic accident. These situations are the result of a lack of coordination between the agencies involved. Additionally, the lack of full practice of the rules and methods to be followed reinforces the upward chart. Risk factors include speeding, drink driving, distracted driving, poor infrastructure, unsafe vehicles,breaking the rules, and many others. A system is therefore required that can ideally coordinate the many steps that need to be taken to ensure rapid response at the scene of an accident. According to  research, such detection systems use various technologies such as deep learning methods and machine learning.Approaches et al. All vehicles are covered by these detection systems, and further technologies  are being investigated. Also we recognizing multiple vehicle types on road. Or the recognition we use one algorithm that is You Only Look Once (YOLOv3 and YOLOv4) is applied for traffic and surveillance applications. A neural community consists of an input with at least one hidden layer and an output. Car, truck, character and bike captured for the duration of RGB and grayscale images. The facts set consists (image and video) of variable lighting. Variants of the YOLO mannequin  such as YOLOv3 are carried out for photograph and YOLOv4 for video dataset. The consequences received exhibit that the algorithm efficiently detects  objects  with an precise accuracy. 

EXISTING SYSTEM:-

 In the beyond decades, the sphere of image control has grown vastly. This has been taken away with the aid of using means: 1)the complete use of images in percent applications, joined with 2) updates withinside the size, speed and value Manuscripts. The sufficiency of reducing area PCs and associated signal orchestrating headways. Picture handling has discovered a simple development in shrewd, cutting-edge, area and authorities applications. Various systems in recent times may be displaced with the aid of using image overseeing change structures that carry out higher than the beyond systems. SDCS gadget is amongst those systems that can claim the normal radars as invalid. This is ideal financially sharp gadget over cutting-edge ones. SDCS shape may be joined with Automatic Number Plate Recognition (ANPR) gadget to form a complete scale radar shape. ANPR shape is a mass reputation method that uses optical man or woman confirmation on photographs to analyze the imprints on vehicles. The makers present the important thing steps toward shape up the Speed Detection Radar. Here makers present any other speculation in factor ID gadget, which is “bendy established order subtraction” because it proofs that it isn’t touchy to startling enlightening changes. Another component is regarded right here regarding cope with following with the aid of using making “item following blueprints”

PROPOSED SYSTEM:-

The proposed approach is based on a visual and temporal feature extractor. The PYTORCH architecture is used in the first step of the model (previously trained on the accident data set). That is, all output cells (convolution layers) were used, leading to the removal of the multilayer perceptron at the end of this design. This is to use the top portion of the model as a visual feature extractor only. However, several tests have shown that the pre-trained model does not distinguish between a stationary car and a vehicle involved in a traffic accident. As a result, the image dataset was used for training.to optimize the weights of this pre-trained network. During this process all the weights of the initial layers of the architecture were frozen and only the weights of the final PyTorch convolution cell were changed. Several tests were performed to fine-tune the Feature Extractor. If the accident is detected, the system notifies the first responders, i.e. the local rescue services. Which directly contributes to reducing response time and saving lives. Because of the difference between a few minutes and even seconds cost a life. because this system

 In our project, we also recognizing multiple vehicle types on road. Or the recognition we use one algorithm that is You Only Look Once (YOLOv3 and YOLOv4) is applied for traffic and surveillance applications. A neural community consists of an input with at least one hidden layer and an output. Car, truck, character and bike captured for the duration of RGB and grayscale images. The facts set consists (image and video) of variable lighting. Variants of the YOLO mannequin  such as YOLOv3 are carried out for photograph and YOLOv4 for video dataset. The consequences received exhibit that the algorithm efficiently detects  objects  with an precise accuracy 

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.
  
  
• Accident Detection:

The implementation of this system is divided into several modules. The system mainly consists of basic modules, the first  is car accident detection, which is achieved by using any high-definition camera, such as CCTV, etc. and then we  use PyTorch to detect crashes. The next module, vehicle detection and classification, was developed using Yolo. Our third module is Velocity Calculation, developed with OpenCV by manipulating and calculating image pixels. Object counting, the TensorFlow is used in this project as the basis for object counting. Finally, the warning system is activated when the vehicle exceeds a certain threshold.
           

ADVANTAGES OF THE PROJECT:-

• Automatic alert on crossing speed limit
• Immediate help can be provided
• Nearest hospitals can be located
• Accident rate can be reduced
• Parents monitor their child driving

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.
• Python
• Anaconda, jupyter notebook, spyder