You are on page 1of 7

DRIVER’S DROWSINESS DETECTION

SYSTEM

Anushka Aman Kumar Arjun Kuamr


Comp Science & Engg. Comp Science & Engg. Comp Science & Engg
Chandigarh University Chandigarh University Chandigarh University
Mohali, Punjab Mohali, Punjab Mohali, Punjab

Pankaj Kumar
Comp Science & Engg
Chandigarh University
Mohali, Punjab

Abstract- In latest years, there has to huge returns for the investors as well as
been a huge growth in the economy, and for the traders. It has been proved that stock
more people have begin to put money forecasting is more complicated than
into the stocks. Predicting stock prices is random, showcasing that the changes in
a difficult and crucial work to perform stock prices can be anticipated by careful
research in the economical field. examination of the history. Machine
Estimating the changes precisely of Learning is a tool to effectively perform
every price of stock can decrease the these operations over the relevant data. It
possibility of losses of captalists. helps in predicting the current value of the
Therefore, efficiently increasing the market that is nearly equal to carrying
return of the respective investments. value and thus providing more accuracy.
This paper aims to propose a model that
predicts the stock prices. This model is
based on Facebook prophet machine
Introduction of Machine Learning to the
learning model that uses additive
regression to estimate the stock prices. stock market has been necessitated more
The prophet model can effectively scope for the research. The data that is used
estimate the prices of stocks over a long for training the model in Machine Learning
time period with rational outcomes. is the most crucial part. Even minor
changes in the dataset can lead to more
Keywords-Machine Learning , Stock Price significant change in the outcome. The data
Prediction, Facebook Prophet, Additive that has been used in this project is taken
regression, Economical field. from yfinance which is a python api
available on Yahoo Finance site. This
1. INTRODUCTION database consists of five variations namely,
Stock Market is the space where assets can high, low, volume, open and close. Volume
be bought, selled and passed around. Proper is termed as the quantity of shares that is
stock estimation and forecasting can govern passed from one owner to another over a
certain period of time. The other four
variants are different stock bid prices at different 2022 Mangesh V. This research uses
periods of times. Estimating stock prices is an volatile Dole, Vishal N. Facebook Prophet to
and non-linear task. It is very difficult to predict stock Sonawane, forecast future stock
Sharique S. market ratings and
values as they are influenced by a variety of factors,
Farooqui, examine how those prices
such as indivisual’s performance, economic and Ehtesham N. will compare to those of
political changes. This ML model can used to Ansari, Prof. earlier stock markets2
maximize profits and minimizes the loss. Algat Y.S2

2. LITERATURE SURVEY
YEAR AUTHOR DESCRIPTION
2021 Nandigama This research also uses
mani ARIMA and Facebook
2022 Yash Saxena, This research uses
Srikara Prophet to inspect the
Ms. Indervati, Facebook Prophet and
Yaswanth, behavior of future stocks Ms. Garima ARIMA models to predict
Kavya and examines the pattern Rathi1 future stock market
Garlapati of stocks2. ratings. The study aims to
make predictions easier
and more realistic1.

2023 Sasmitoh Stocks are an investment


Rahmad instrument that is
Riady starting to be in great
demand by
2022 Manish This research uses the public today.
Mahajan Facebook Prophet to However, stock prices
Bharti Joshi forecast future stock are fluctuating, making
Aditya market ratings and people feel
Janghale examine how those
prices will compare to
those of earlier stock
markets2.
2023 Harmanjeet This research presents video feed from a camera placed inside the
Singh & an architecture based on vehicle, focused on the driver's face. The
Manisha a time series model, video frames are then fed into the trained
Malhotra such as Facebook model, which analyses the facial
Prophet, Light Gradient expressions and provides a prediction of the
Boost Machine (GBM), driver's drowsiness level.
and Autoregressive
Integrated Moving
To ensure real-time performance, the model is
Average with
typically optimized for speed and efficiency. This
Explanatory Variable
(ARIMAX) tocan involve techniques like model compression,
accurately predict stockquantization, or using specialized hardware
prices4. accelerators.

The output of the drowsiness detection


model can be used in various ways. For
example, it can trigger an alert system to
notify the driver when drowsiness is
detected, such as beeping an alarm or
3. METHODOLOGY
vibrating the seat. This can help prevent
The first step is to gather a large dataset of accidents usually caused by driver fatigue.
driver's facial images. These images should
cover a wide range of individuals, facial It's worth noting that the accuracy and
expressions, and lighting conditions to performance of the drowsiness detection
ensure the model's robustness. The dataset system depend on various factors, including
should also include labelled annotations the quality of the dataset, the architecture of
indicating whether the driver is drowsy or the deep learning model, and the real-world
alert. conditions in which it is deployed.
Next, a deep learning model, such as a
convolutional neural network (CNN), is 4. SPECIFICATION
trained on this dataset. The CNN learns to
extract relevant features from the facial Deep Learning- Deep Learning is sub-part
images that can help distinguish between or we can say subset of Machine Learning,
drowsy and alert states. The network is it is very useful in solving complex
trained using a combination of forward and problems and it involves much use of
backward propagation, adjusting the neural networks to model. Deep Learning
weights and biases to minimize the has gained a significant amount of success
predictionerror. in numerous fields, encompassing natural
language processing, image recognition,
During training, the network learns to speech recognition, and recommendation
recognize patterns and features that are system. Some of the voguish deep learning
indicative of drowsiness, such as drooping architectonics encompass Convolutional
eyelids, eye closure, yawning, and changes Neural Networks (CNNs), Recurrent
in facial expressions. By analysing these Neural Networks (RNNs), Deep Belief
patterns, the model can make accurate Networks (DBNs).
predictions about the driver's drowsiness
level. Additive Regression- It is a type of
regression model with no parameters
Once the model is trained, it can be which was proposed by Jerome H.
deployed for real-time drowsiness Friedman and Werner Stuetzle in 1981. In
detection. This involves capturing live this linear output variable depends linearly
on distributed functions of unknown values.
Stock Price Prediction- This system 6. RESULT AND DISCUSSION
involves estimating the buying and selling
prices of different assets and stocks. It also The system keeps detecting the face until it
helps in approximation of future prices of detects the driver is being drowsy. Once the
the existing stocks in the market on the face is detected the system scores the driver
basis of past data and various other on the basis of his/her eyes and its
indicators of the market. movements.
If the Score is greater than 15 the alarm
5. WORKING starts beeping to warn and awaken the
driver and a message is displayed ‘closed’.
If the Score is less than 0 then the score is
reinitialized to 0 and the message is
displayed as ‘open’.
The ‘open’ message is also displayed when
the Score is less then 15. In case, the
message being displayed is ‘open’ the
alarm does not beep or stops beeping.

Fig 1.2

Fig 1.1

Above figure, in the form of flowchart,


shows the working of the system. Firstly, it
takes the database as training dataset which
is basically the historical trends of stock
price. Further it decides the future close
price of the stock using Prophet algorithm
and gives the plot predicted graph.

Fig 1.3

Fig 1.2 shows a ‘closed’ message having


Score as 170 which is greater than 15, thus the [2] Manish Mahajan, Bharti Joshi,
alarm starts to beep. Aditya Janghale (2022). Research on
forecasting future stock market ratings
using Facebook Prophet.

[3] Mangesh V. Dole, Vishal N.


Sonawane, Sharique S. Farooqui,
Ehtesham N. Ansari, Prof. Algat Y.S
(2022). Research on forecasting future
stock market ratings using Facebook
Prophet.

[4] Yash Saxena, Ms. Indervati, Ms.


Garima Rathi (2022). Research on
predicting future stock market ratings
using Facebook Prophet and ARIMA
models.

Fig 1.4
[5] Sasmitoh Rahmad Riady (2023).
Research on the demand for stocks as
7. CONCLUSION an investment instrument and the
fluctuation of stock prices.
The model built, successfully, gives plot
predicted graph on the basis of historical
data. The paper is successful in [6] Harmanjeet Singh, Manisha
demonstrating that a model can be built Malhotra (2023). Research on
using Facebook Prophet algorithm that uses predicting stock prices using a time
additive regression in order to draw the plot series model, such as Facebook
graph, with high accuracy and low Prophet, Light Gradient Boost
feasibility. It studies the trends, future and Machine (GBM), and Autoregressive
history of the stock prices and gives the Integrated Moving Average with
required output on the basis of that. Thus, Explanatory Variable (ARIMAX).
fulfilling the motto of reducing losses
and risks, thus maximizing profits for the
[7] Wu, Jimmy Ming-Tai & Li,
investors as well as for the traders. It
Zhongcui & Herencsar, Norbert & Vo,
analyses and gives a proper speculations for
Bay & Lin, Chun-Wei."A graph-based
the consumers to invest on the particular
stock on on the basis of analysed timeline. CNN-LSTM stock price prediction
algorithm with leading indicators" in
Multimedia Systems.February
8. REFERENCES
2021,DOI:10.1007/s00530-021-
[1] Nandigama Mani Srikara Yaswanth, 00758-w
Kavya Garlapati (2021). Research on the
behavior of future stocks using ARIMA and
Facebook Prophet. [8] Mehtab,Abhishek Dutta."Stock
Price Prediction Using Machine
Learning and LSTM-Based Deep
Learning Models" in Second
Symposium on Machine Learning and
Metaheuristics Algorithms.October
2020,DOI:10.13140/RG.2.2.23846.34880

[9] Mislinski, Jill (3 March 2020).


"Market Cap to GDP: An Updated Look at
the Buffett Valuation Indicator".
www.advisorperspectives.com. Archived
from the original on 14 March 2020. it
is probably the best single measure of
where valuations stand at any given
moment”.

[10] Hong, Tianyi, and Huabiao Qin.


"Drivers drowsiness detection in embedded
system." In 2007 IEEE International
Conference on Vehicular Electronics and
Safety, pp. 1-5. IEEE, 2007.

[10] Thawornwong, S, Enke, D.


Forecasting Stock Returns with Artificial
Neural Networks, Chap. 3. In Neural
Networks in Business Forecasting,
Editor: Zhang, G.P. IRM Press, 2004.

[11] Niloy, Amit Raha, Atiqul Islam


Chowdhury, and Nusrat Sharmin. "A brief
review on different Driver's drowsiness
detection techniques." International Journal
of Image, Graphics and Signal Processing
10, no. 3 (2020): 41.

You might also like