The project will introduce and apply the Object Detection and Tracking System(ODTS) along with the well_known faster RCNN.

1. A DEEP LEARNING ALGORITHM FOR AUTOMATIC DETECTION OF
UNEXPECTED ACCIDENTS UNDER BAD CCTV MONITORING
CONDITIONS
M Sai Sree
BY

2.  Introduction
 Abstract
 Problem Statement
 Existing System
 Proposing System
 System Requirements
 References

3. As the urban population rises and the number of motor vehicles increases, accidents becoming a
major concern. Every year around 1.35 million people are cut off due to numerous crashes.
In addition to accidents , these accidents cause major traffic to avoid these situations the early
detection of accidents have to be done.
For Control systems, monitoring of surveillance camera is difficult to detect unexpected events like
vehicle accidents ,fire accidents and wrong way driving due to various conditions like weather,
lightening conditions etc.
This project is to create an accident detection method that can acquire moving details of target
objects by integrating an object tracking algorithm with deep learning based object detection
mechanism.

4.  In this project, Object detection and tracking system (ODTS) in combination with well-Known deep learning
network, Faster regional convolution neural network (faster R-CNN), for object detection and conventional object
tracking algorithm will be introduced and applied for automatic detection and monitoring of unexpected events on
CCTVs.
 Here it detects
 Wrong-way driving(WWD)
 Person
 Fire
 Car
 The ODTS was tested using accident videos which including each accident. As a result it can detect the accident
with seconds.

5. To develop a project that uses the techniques of deep learning to detect unpredictable
accidents under poor CCTV monitoring circumstances using various algorithms
To detect the object and conventional tracking of object using Faster R-CNN which is most
popular deep learning network.
Driving in a road tunnel is dangerous because it is inadequate space to evacuate compared to
a general highway.
Therefore, drivers should be informed as soon as possible when an emergency occurs in the
tunnel.

6. In [1] , deep learning-based algorithm system in combination with CNN and SVM was
developed to monitor moving vehicles on urban roads or highways by satellite.
According to [2], unsupervised motion clustering detect moving objects with no previous
knowledge of their visual presence, location or shape.
In [3], the possibility of exploitation of deep neural features for robust vehicle detection is
examined using DNN.
In [4], Predicting trajectories of traffic participant in traffic as well as their future location.
Use of a Convolutional LSTM Auto- Encoder.

7. This system utilizes a faster RCNN algorithm for object detection and a SORT for ID assignment
and object tracking.
A deep learning model of Faster R-CNN was used training. And this model was based on a model
that learns image datasets that include some accidents.
This system utilizes BBox obtained by object detection on videos or images.

8. HARDWARE REQUIREMENTS
• System : Intel Core i7
• RAM : 12.0 GB
• HardDisk : 98 GB
• Monitor : 15 “ LED
SOFTWARE REQUIREMENTS
• Operating System: Windows 10
• Coding Language: Python
• Packages : Opencv, cmake
• Tool : Command prompt

9. 1. E. S. Lee, W. Choi, D. Kum, “Bird’s eye view localization of surrounding vehicles :Longitudinal and lateral
distance estimation with partial appearance,” Robotics and Autonomous Systems, 2019, vol.112, pp. 178-189.
2. L. Cao, Q. Jiang, M. Cheng, C. Wang, “Robust vehicle detection by combining deep features with exemplar
classification,” Neurocomputing,2016, vol. 215, pp. 225-231.
3. A. Bewley, V. Guizilini, F. Ramos and B. Upcroft, "Online self-supervised multi-instance segmentation of
dynamic objects," 2014 IEEE International Conference on Robotics and Automation (ICRA), Hong Kong, China,
2014, pp. 1296-1303
4. Yao, Y., Xu, M., Wang, Y., Crandall, D.J. and Atkins, E.M., 2019. Unsupervised traffic accident detection in first-
person videos. arXiv preprintar Xiv: 1903.00618.