A Deep Learning Based Vehicle Detection with Emergency Notification

DOI : 10.17577/IJERTCONV12IS03096

Download Full-Text PDF Cite this Publication

Text Only Version

A Deep Learning Based Vehicle Detection with Emergency Notification

Anandhan KC Assistant Professor

Department of Electrical and Electronics Engineering,

Shree Venkateshwara Hi-Tech Engineering College, Gobi ,India anandhan444@gmail.com

Ajay sooriya Department of Electrical and

Electronics Engineering, Shree Venkateshwara Hi-Tech Engineering College, Gobi, India


Arumugam. T Department of Electrical and

Electronics Engineering, ShreeVenkateshwaraHi-Tech Engineering College, Gobi India, arumugam2003eee@gmail.com

Vintha .S Department of Electrical and

ElectronicsEngineering, ShreeVenkateshwaraHi-Tech Engineering College, Gobi, India vinithavini960@gmail.com

Abstract Vehicle riding is now safer because to the type of protective helmet that rider wear. Ensuring the safety of the rider is the primary goal here. Advanced features like fall detection, alcohol detection, accident identification, location monitoring, and hands-free device use can be used to implement this. As a result, it becomes more than just a clever helmet. Helmet wear is required; the ignition switch cannot operate without it. A transmitter and receiver can communicate wirelessly by using an RF module as a connection. The ignition locks automatically and sends a message to the registered number with the rider's current position if the rider is intoxicated. Should an accident occur, it will transmit communicate via GSM and provide location data via GPS module. Fall detection, which sends a message if the rider falls off the bike, is the project's unique feature. Smart helmet technology can possibly save lives and lessen the strain on the healthcare system by lowering the likelihood of accidents and injuries on the road through features like alcohol detection, accident alarms, and smart vehicle detection. We may anticipate seeing even more sophisticated and practical smart car designs in the future as technology develops and advances.


Consistently, roughly 1.35 million individuals bite the dust on account of street mishaps. In view of these mishaps, more than 20 to 50 million individuals experience the ill effects of non-lethal wounds, and many become handicapped from these wounds. There is a need to focus on make youngsters more mindful of street wellbeing issues. The test is to decrease the quantity of mishaps. For which, it's smarter to act before a mishap happens. However, some of the time it's not in our grasp. Thus, many harmed lose their lives for not getting educated ahead of time. It is a goliath misfortune and working out the social expense would uncover how enormous it is an ascent in the quantity of mishaps and the quantity of individuals winding up dead or harmed in those mishaps is positively a main issue. Nonetheless, it is the obligation of each and every individual particularly the young people who go onto the streets that they should be protected and observe the traffic guidelines. Helmeted bikers experienced essentially less and less serious head and neck wounds than unharmed riders. Individuals who were harmed had head and neck brought about by the wellbeing caps, every one of these wounds was somewhat minor, and head protectors shielded them from possibly deadly wounds. In this manner, caps significantly diminish

the quantity of passing and wounds by cruiser mishaps. The primary commitments of this paper can be brief as follows: a) Fuel robbery b) Over-burden c) Over speed d) Liquor recognition

e) Not wearing cap f) Mishap recognizable proof g) Putting away about the vehicle data set like ended up in a good place, speed of the vehicle, mileage of the vehicle. The primary driver of death in many mishaps that clinical benefits don't show up on time. In remote spots, where the mishaps occur, nobody can be educated to the emergency vehicle. A brilliant protective cap is utilized for ID mishap and forestalling the mishap. There use IR sensors, load sensors for forestalling a mishap. The accelerometer is utilized for identifying any mishap. The accelerometer really takes a look at the slant of the protective cap to decide the event of a mishap. PIC miniature regulator is utilized for handling checking framework sends the area to the concerned power. GPS global positioning framework identifies the exact area data. At the point when mishaps occur, the framework identifies the mishap and sends a mishap area to the closest clinic by the GSM Module.

The start switch won't begin until the rider wearing the cap. Alcoholic driving is beyond the realm of possibilities when wore a savvy cap. In the event that the rider is alcoholic, the bicycle won't begin and the send message to the enrolled number. The clients can without much of a stretch recognize the fuel robbery in the bicycle. By involving level transmitter in the power source of the tank distinguish the fuel burglary. Because of over speed numerous mishaps happen. To keep away from over speed, speed limit sensor introduced in the spinning engine, the speed getting limit message ship off rider through the signal sounder to the rider. The data like everyday voyaged places, fuel utilization of vehicle, vehicle mileage and speed of the vehicle can be put away in the cloud Per vehicle mile voyaged, motorcyclists are multiple times more probable than a traveler in a vehicle to kick the bucket in an accident. Also, the greater part of cruiser fatalities in 2013 were unhealed riders. Most bike mishaps that outcome in death are brought about by a head injury.

Saving a daily existence and forestall a cerebrum injury rider ought to wear a protective cap. Serious wounds are enormously decreased both in seriousness and in recurrence by the basic demonstration of wearing a head protector. Presently mishaps on streets have turned into a serious worry for every one of the large number of individuals particularly young people are either winding up dead or harmed consistently. The most recent gauge by the World Wellbeing Association additionally shows that internationally world street traffic wounds are the main source of death among youngsters matured 15 to 29 years. Every

year, around 3.4 lake young people in this age bunch passed on in mishaps as per the report ready by theservices transport research wing. The absolute number of street crashes has expanded barely from 3.86 lakes in 2013 to 4.50 laky in 2014. The quantity of fatalities has likewise gone up by around 1.5 percent.

This shows that there is a need to focus on make youngsters more mindful of street security issues. It is a mammoth misfortune and working out the social expense would uncover how huge it is an ascent in the quantity of mishaps and the quantity of individuals winding up dead or harmed in those mishaps is unquestionably a central issue. Notwithstanding, it is the obligation of each and every individual particularly the young people who go onto the streets that they must be protected and keep the traffic guidelines. The primary commitments of this current work can be brief as follows: a) forestalling mishap, b) lessening mishap causality) recognizing the mishap and d) fostering the mishap related traffic the executives framework.

In this paper, we propose a model of a savvy cap for mishap recognizable proof and forestalling the mishap. There use IR sensors, gas sensors, and burden sensors for forestalling a mishap. The 3-pivot accelerometer is utilized for identifying any mishap. Adriano is utilized for handling sensors information and building a correspondence framework among sensors and portable applications. The versatile application is associted with a focal checking framework 9 and authority can screen each client mishap history. At the point when any mishaps happen, the mishap area ships off the checking data set, and the observing framework sends the area to the closest emergency clinic and police focus. Adriano recognizes an accident if there is a sudden change in acceleration.

Adriano takes two minutes for a touch sensor response following an accident. For the little accident, a touch sensor was used. In the event of an accident, the rider's touch sensor will register the event as moderate, meaning no message will be sent to the emergency contact number. The emergency contact number and the accident's current location are automatically sent to the database if the rider is unable to touch the sensor. Through the use of the phone's integrated GPS (Global Positioning System) module, the mobile application determines the user's present location. Moreover, the emergency contact number receives the position information via GSM (Global System for Mobile communication).

Creating intelligent objects with the wide range of talents we see in people was the first goal of artificial intelligence. The tone of many influential works makes this motivation very evident, even though none of the field's early projects succeeded in achieving this noble objective. This goal remained significant in many circles when we became professors in the mid-1980s, even though it was a key premise of many researchers in the field when we started graduate school at Carnegie Mellon University in the mid- 1970s. The prevailing consensus at the time was that AI was a single field with a shared set of objectives. That was beginning to alter by the late 1980s. Sub disciplines like as planning, knowledge representation, and machine learning started to separate off. geometric intricacy. The latest machine learning department at Carnegie Mellon has fully developed this tendency, requiring its graduate students to complete numerous statistics courses but not a single artificial intelligence course. There has been a change in the difficulty of the issues addressed along with this affiliation change. Every subfield has established a clear set of objectives for which it tracks advancement. In machine

supervised learning for classification and learning from delayed reward for reactive control; similar issues exist in other subfields. They don't reveal much about the nature of intelligence, despite their benefits for assessment purposes and, occasionally, their practical utility (for example, corporate data mining). Additionally, interactive computer games are becoming more and more popular, making them a viable field of study for cognitive system researchers. These games offer realistic environments that enable one to address many of the integration challenges that occur in designing.

AI systems that are capable of functioning at the human level, as noted by Laird and van Lent (2000). They also bypass the hardware constraints related to robotics research and are quite attractive to many students. This realization is expanded upon in the recently launched conference on AI and Interactive Digital Entertainment (http://www.aiide.org/), which is unquestionably a step in the right direction. There are numerous benefits associated with intelligent tutoring systems for the study of integrated cognitive systems. They need to be able to deduce students' knowledge states in addition to having knowledge of the target domain. The goal of the large field of computer science known as artificial intelligence (AI) is to create intelligent machines that can carry out tasks that normally call for human intelligence. To put it another way, its goal is to create computers. clever and intelligent by granting them the capacity to learn and think like computers or software, i.e., to think and operate similarly to humans. Philosophically speaking, AI has the ability to govern the vast network of interconnected people, companies, states, and nations in a way that is advantageous to everybody while also enabling people to lead more fulfilling lives with less effort. Making computers and other devices capable of cognitive tasks like problem-solving, judgment, perception, and understanding human speech is, thus, the main objective of artificial intelligence (AI).

Building automated, intelligent, and smart systems to meet today's needs is therefore dependent on AI-based modeling, which has emerged as the next significant technical milestone, impacting the Despite the vastness of the field of "artifcial intelligence," we primarily concentrate on prospective methods for resolving practical problems, the outcomes of which are applied to the development of automated, intelligent, and intelligent systems across a range of markets.We divide up different AI approaches into ten categories so that we can create models based on AI: machine learning; deep learning and neural networks; data mining, knowledge discovery, and advanced analytics; rule-based modeling and decision-making; fuzzy logic-based approach knowledge representation, expert system modeling, and uncertainty reasoning; case-based reasoning; text mining and natural language processing; visual analytics, computer vision, and pattern recognition; hybridization, searching, and optimization.

The "Potential AI techniques" in briefIn this work, we present a thorough understanding of "AI-based modeling," which can be crucial in achieving automation, intelligent, and smart systems that meet modern needs, based on the significance and potential of AI methodologies. In order to fulfill the demands of the Fourth Industrial Revolution, the main aim is to clarify the fundamentals of various AI techniques and how they might be used to the development of computers and decision-making. Since artificial intelligence techniques are the basis for developing automated, intelligent, and smart systems. The aim of this paper is to offer a basic reference for academics.

This paper's primary contributions are consequently enumerated as follows:

leaVrnoliunmg,e 1f2o,r Isesuxeam03ple, the focus is nearly entiPruelbylishoend by,


ISSN: 2278-0181

  • To define the parameters of our investigation with respect to automation, intelligent and smart computing, and decision- making in light of current practical requirements.

  • To study the theme of artificial intelligence's power in computing and decision-making while addressing a range of issues in the current Fourth Industrial Revolution. This involves investigating different forms of AI, such as analytical, functional, interactive, textual, and visual AI.

  • To give a thorough overview of AI approaches that can be used to create an AI-based model that will improve the functionality and intelligence of a real-world application.

  • To help academics and developers extend their perspectives on artificial intelligence by talking about how AI-based solutions can be applied in a variety of real-world application domains.


    The Fourth Industrial Revolution, or Industry 4.0, as it is commonly known, is currently upon us and heralds a new era of technological innovation, especially in the area of artificial intelligence (AI). Following the Third Industrial Revolution, which was triggered by the Internet and mobile devices, Industry 4.0 is currently being ushered in by data- driven AI technology. Generally speaking, "Industry 4.0" describes the current trend of using contemporary technology to automate procedures and exchange data. The term "Industry 4.0" refers, broadly speaking, to the current trend of industrial technology automation and data sharing, which encompasses the creation of cloud computing, cognitive computing, cyber-physical systems, the Internet of Things, nd Industrial AI has the potential to improve current products and services, making them more efficient, dependable, and secure. To make driving safer, computer vision is used, for instance, in the automotive sector to prevent crashes and enable vehicles to stay in their lanes.

    The most powerful countries in the world are rushing to increase their investments in artificial intelligence (AI). Likewise, the biggest and most. Understanding and performing intelligent tasks, such reasoning, learning new skills, and adjusting to novel situations and obstacles, are at the heart of artificial intelligence (AI). AI is therefore regarded as a field of study and engineering that focuses on modeling a variety of problems and activities related to the field of human intelligence. But creating an effective AI model is a difficult endeavor because real-world scenarios and data are dynamic and diverse.many forms of artificial intelligence (AI), such as analytical, functional, interactive, textual, and visual, to address a variety of problems in the current Fourth Industrial Revolution and comprehend the concept of AI's power Analytical AI: Analytics is the general term for the process of finding, deciphering, and sharing significant data patterns. Analytical AI thus seeks to support data-driven decision-making by identifying novel insights, patterns, correlations, or dependencies in data.

    As a result, it becomes a fundamental component of AI in the context of modern business intelligence, capable of offering insights to an organization and producing ideas or recommendations via its analytical processing power. An analytical AI model can be created using a variety of deep learning [80] and machine learning [81] techniques to address a specific real-world issue. For example, a data-driven analytical model can be used to evaluate business risk

    .Functional artificial intelligence (AI) functions similarly to analytical AI in that it examines vast amounts of data for because it examines vast amounts of data for patterns and connections, functional AI functions similarly to analytical AI. Conversely, functional AI does not provide

    recommendations; rather, it carries out tasks. beneficial in robots and Internet of Things applications to make quick decisions.

    Fig 1: Vibration Sensor

    Intelligent automation that is both efficient and dynamic in communication is known as interactive artificial intelligence (AI). This type of automation is widely used in many areas of daily life, especially in the business world. An interactive AI model, for example, might be helpful in the development of chatbots and intelligent personal assistants. An interactive AI model can be constructed using a range of methods, including machine learning, frequent pattern mining, reasoning, and AI heuristic search.Textual AI: This category usually includes text analytics or natural language processing, which gives organizations access to features like machine translation, text recognition, speech-to-text conversion, and content creation. To maintain an internal corporate knowledge library and offer pertinent services, such as responding to customer inquiries, an organization can, for example, utilize textual AI.


    One of the most exciting areas of artificial intelligence (AI) is machine learning (ML), which is basically the study of computer algorithms that automate the creation of analytical models [81]. In order to find intriguing data patterns or predict behavior, machine learning (ML) models are frequently composed of a collection of guidelines, protocols, or complex "transfer functions" .

    Machine learning, commonly referred to as predictive analytics, is the process of using data to forecast specific unknowns in the future. It is used to handle a variety of real- world business problems, such as estimating company risk. A machine learning-based predictive model's overall framework is shown in Fig. 3. In phase 1, the model is trained using historical data, and the result is produced for new test data. Additional methods of instruction Because it uses both labeled and unlabeled data to train a model, semi-supervised learning can be thought of as a hybridization of the supervised and unsupervised tasks mentioned above. When data needs to be labeled automatically without human intervention, it could be useful for enhancing model performance.

    A semi-supervised learning model could be helpful, for example, in the classification of texts or Internet information. Another kind of machine learning training technique that encourages desired behavior while penalizing undesirable behavior is called reinforcement learning. In general, an environment-driven method to reinforcement learning involves an agent's ability to perceive and interpret its surroundings, act, and learn through trial and error. Hybrid learning with deep networks The versatility of generative models allows them to learn from both labeled and unlabeled input. On the other hand, discriminative models perform better in supervised tasks than

    their generative counterparts, but they are not able to learn from unlabeled data. A paradigm for simultaneously training deep generative and discriminative models serves as the driving force behind hybrid networks. Hybrid deep learning models consist of many (two or more) deep basic learning models, the discriminative or generative deep learning model described above serving as the basic model. For example, addressing real-world problems may benefit from the combination of a generative and discriminative model followed by a non-deep learning classifier.

    Fig 2. Microcontroller

    However, the majority of information probably goes toward recalling visual cues and other relatively inefficient uses. Therefore, given the limitations and volatility of natural intellect, it is possible that the world will increasingly rely on computers to function properly.Artificial intelligence (AI) is rapidly becoming as a fundamental component of computer science.NetApp recognizes the importance of data management, access, and control in its role as the hybrid cloud's data authority. Unified data management across numerous hyper scale clouds, data centers, and edge devices is made possible via the Net App data fabric. Organizations of all sizes may expedite vital applications, obtain data visibility, simplify data security, and boost operational agility with the help of the data fabric.

    We may choose to stick with artificial intelligence given its features and broad range of applications. Is the world of the future turning artificial given the advancement of AI? Because it is an established, mature paradigm, biological intelligence cannot be changed, but the new paradigm of non-biological computing and intelligence is expanding at an exponential rate. It is likely that the human brain can store approximately ten thousand million binary digits inmemory. Using IoT devices and aggregation points, ONTAP Select software facilitates effective data collecting at Utilizing Cloud Volumes can help. Address learning actually refers to the process of determining the correct address (MAC or hardware address) and then using that specific address to facilitate additional device connection.

    The problem is that a switch's filter table is empty when it is first turned on, meaning that no information is accessible on which device is attached to which port or switch interface. So, how does the switch update the blank filter table with the device's address? Alright! Anytime a device begins to communicate using both the recipient's and its own logical address (IP address, which is previously known), The switch then stores the frame's source address in the filter table during the initial transmission process, which enables the switch to remember which interface the sending device is located on in the future. The frame contains the IP address of the source and destination, the MAC address of the source, and other information. The sender#39;s MAC address

    has now been added to the filter table; nevertheless, the recipient's MAC address is still unknown; only its IP address is known. How will the received frame at the switch be routed to the appropriate device, then? Broadcasting now steps in to save the day.

    Fig 3.Motor Driver

    A deeper knowledge of a business's evolution can be gained by analyzing historical data through descriptive analytics. In order to answer the question, "What happened in the past?" descriptive analytics describes historical data, including marketing strategies, social media activity, sales and operational statistics, and so forth.

    Diagnostic analytics is an advanced form of analytics that investigates data or content to determine the "why" behind an event. Aiding in the identification of the issue's primary cause is the goal of diagnostic analytics.

    Fig 4.Simulation of accident alert system

    Predictive analytics: This branch of advanced analytics usually focuses on data analysis to provide a solution to the question, "What next?" With that in mind, the main goal of predictive analytics is to recognize and, for the most part, provide a high degree. In summary, the goals of both diagnostic and descriptive analytics are to understand what happened in the past and why it happened. Prescriptive and predictive analytics use past data to forecast future events and recommend appropriate course.


Up till now, we have had a brief discussion on artificial intelligence. A few of its tenets, uses, accomplishments, etc., have been covered.Achieving the majority of issues or activities that humans are unable to perform directly is the ultimate goal of AI research institutes and scientists. Undoubtedly, advancements in computer science will fundamentally alter the global landscape. Currently, the onus is on the upper echelons of engineering to propel this area forward.


  1. Anandhan K C, Jayakumar T, Ponmurugan P, Ragul S, Karthik B, Neelam Sanjeev Kumar, Rajkumar D Review Of Hybrid Wind-Solar Pv Technology In The Generation Of Electricity IEEE Conference.

  2. Brachman, R., & Lemnios, Z. (2002). DARPAs new cognitive systems vision. Computing Research News, 14,

  3. Cassimatis, N. L., Mueller, E. K., & Winston, P. H. (2006). Achieving human-level intelligence through integrated systems and research. AI Magazine, 27, 12 14.

  4. Laird, J. E., & van Lent, M. (2000). Interactive computer games: Human-level AIs killer application. Proceedings of the Seventeenth National Conference on Artificial Intelligence (pp. 11711178). Austin, TX: AAAI Press.

  5. Langley, P., Laird, J. E., & Rogers, S. (in press). Cognitive architectures: Research issues and challenges. Cognitive Systems Research.

  6. Tambe, M., Johnson, W. L., Jones, R. M., Koss, F., Laird, J. E., Rosenbloom, P. S., & Schwamb, K. B. (1995). Intelligent agents for interactive simulation environments. AI Magazine, 16.

  7. Trafton, J. G., Cassimatis, N. L., Bugajska, M., Brock, D., Mintz, F., & Schultz, A. (2005). Enabling effective human-robot interaction using perspectivetaking in robots. IEEE Transactions on Systems, Man and Cybernetics, 25, 460470.

  8. Ester M, Kriegel H-P, Sander J, Xiaowei X, et al. A density-based algorithm for discovering clusters in large spatial databases with noise. In Kdd. 1996;96:22631.

  9. Fatima A, Maurya R, Dutta MK, Burget R, Masek J. Android malware detection using genetic algorithm based optimized feature selection and machine learning. In: 2019 42nd international conference on telecommunications and signal processing (TSP), IEEE; 2019. p. 220223.

  10. Goel D, Pahal N, Jain P, Chaudhury S. An ontology- driven context aware framework for smart trafc monitoring. In: 2017 IEEE region 10 symposium (TENSYMP), IEEE; 2017. p. 15.

  11. González-Briones A, Prieto J, De La Prieta F, Herrera-Viedma E, Corchado JM. Energy optimization using a case-based reasoning strategy. Sensors. 2018;18(3):865.