International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 05 | May 2023
p-ISSN: 2395-0072
www.irjet.net
ACCIDENT DETECTION USING BiLSTM Sreechandana Salvaji1, Sai Teja Asuri2, Divya Sree Nemmikanti3, Akash Raj4, Mrs. P. Radhika5 1,2,3,4 Student, Department of CSE, VNR VJIET, Hyderabad, Telangana, India
5Assistant Professor, Dept. of Computer Science and Engineering, VNR VJIET, Telangana, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Accidents have consistently ranked as the
combined virtual technologies with it to create the Intelligent Transport System. In the sphere of transportation, the concept of integrating virtual technologies is innovative, and it is essential for resolving problems in a worldwide context. The traditional method for creating next-generation technology is known as ITS. From ITS implementations, a variety of reimbursements are available. ITS can significantly lower hazards, accident rates, traffic jams, carbon emissions, and air pollution and meanwhile, improving all modes of transportation's traffic flow, transit speeds, and levels of passenger satisfaction. One of ITS's key uses is traffic control. Controlling traffic is becoming a major challenge as the overcrowding issue gets worse. The Video traffic surveillance system is one of the key technologies being used to implement solutions for this problem.
major cause of death in India. More than eighty percent of the fatalities that occur as a result of accidents are not directly attributable to the accident itself; rather, they are the result of victims not receiving prompt assistance. It is possible for an accident victim to be left unattended for a significant amount of time on routes that have very light and quick traffic. The objective is to design a system that is able to determine whether or not an accident has occurred based on the video input received by the system. It is the intention to run each frame of a video through a convolutional neural network and BILSTM models that have been trained to identify video frames as either accident or non-accident frames. The Convolutional Neural Network and the BiLSTM models have been shown to be a method that is both quick and accurate when it comes to identifying photographs. CNNbased image classifiers have attained an accuracy of greater than 95% with fewer datasets, and they require less preprocessing than other image classification techniques. Key Words: BILSTM
2. RELATED WORK [1] As analytical tools in this particular study project, CNN, RNN, and LSTM were employed. Four layers make up the study's architecture: two convolutional layers that help with feature extraction, two layers of long short-term memory (LSTM) units, and a top layer. LSTM is in charge of controlling each video's time dependence (Long Short-Term Memory). Over 80% accuracy in validation is achieved, with the sheer amount of data being one of the main challenges.
Convolutional neural network and
[2] Transfer Learning and Mask R-CNN, which uses a mask RCNN to detect cars, are the main techniques that we employ in this study. The intersection over union (IoU) algorithm is used in order to discover collisions. If used in conjunction with a strong Response system, this model could reduce wait times, speed up procedures, and improve detection accuracy.
1.INTRODUCTION The main goal is to implement a system that can recognize an accident from provided video material. The system is meant to be a tool to assist an accident by promptly identifying an accident and afterward reporting the authorities about it, you can help those in need. The goal is to use cutting edge Deep Learning Algorithms that use BILSTM and Convolutional Neural Networks (CNNs or ConvNet) to analyze frames captured from the video input given to the system in order to identify an accident within seconds of it occurring. We concentrated on installing this technology on highways where there is less congestion and prompt assistance for accident victims is uncommon. Transportation is a legitimate means of taking or carrying items from one location to another. As time goes on, transportation suffers a number of problems, including a high accident rate, traffic jams, air pollution from carbon emissions, and more.
[3] The cornerstone of the accident detection system is provided by the CVIS and machine vision. We designed the YOLO-CA deep neural network model to discover accidents. Deep learning techniques and CAD-CVIS were used to create this model. We use a loss function with dynamic weights and Multi-Scale Feature Fusion (MSFF) to enhance the recognition accuracy of very small objects. When it comes to choosing proposal regions, Fast R-CNN uses the timeconsuming selective research approach. When dealing with very large objects, rapid R-detection CNNs give incorrect positive findings. [4] The results of this study suggest that the best approach to the issue is to apply video analytics techniques. The structure is composed of two distinct components. The first
The transportation industry occasionally struggled with reducing the severity of crash-related injuries in accidents. Since transportation is so complex, researchers have
© 2023, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 297