Pothole Detection Using an Accelerometer and Image Processing

Sunil Sable, Bhushan Shivane, Vaishnavi Shivakumar, Ashlesha Todkar, Prakriti Singh,
Shivraj Kadam, Shreya Tonape
combination of image analysis and sensor data allows
for efficient maintenance planning, improved road
[email protected], safety, [email protected],
[email protected], and timely repairs, leading to enhanced
infrastructure quality.
[email protected], [email protected], [email protected],
[email protected]

Department of Multidisciplinary Engineering

Vishwakarma Institute of Technology, Pune, 411037, Maharashtra, India

Abstract- This paper presents a pothole detection

system that uses an accelerometer and GPS. The II. Literature Review
system detects potholes by measuring the magnitude of
acceleration and triggering an alert when the Pothole detection is a worthy topic of research, several
acceleration exceeds a predefined threshold. The GPS methods are examined by the experts. In this section,
module is utilized to determine and provide the we illustrate various research that has been developed
accurate location of the pothole., which is then sent to to detect potholes and has been addressed in a variety
a remote server. To provide the date, time, longitude,
of ways and systems so far.
and latitude data from a Google Sheet to an AppSheet
app this system provides a pothole detection AppSheet
app. These Appsheets can be accessed by authorized In the paper Pothole Detection System for Monitoring
users to view the location of the potholes detected by Road & Traffic Conditions using IoT, [1] they have
the system. The system was tested on a moving vehicle proposed a way where they have built a pothole sensor
and demonstrated promising results in detecting using IOT, and the roads with potholes will be tracked
potholes accurately.
by the GPS module. The data about roads with
potholes will be tracked and fed to the server. This is
Keywords- Pothole Detection, Accelerometer, Global
Positioning System (GPS) Module, Internet of Things done by placing the sensor on many vehicles. This data
(IoT), accessed, Appsheet. on roads with potholes was tracked with the help of a
mobile application. Thus, any other driver can use that
I. Introduction application to get data about roads with potholes and
Potholes are a major problem for road safety, causing prevent accidents. The pothole detection system was
damage to vehicles and posing a risk to drivers and developed using a Raspberry Pi, and the collected data
passengers. Traditional methods for detecting was transmitted to the server using the Wi-Fi module.
potholes involve manual inspections, which are time- There were accelerometers and gyroscope sensors to
consuming and often inaccurate. This paper presents measure the vehicle’s speed and torque. This data was
a pothole detection system that uses an accelerometer transferred to a Python interpreter to analyze it and
and GPS to automatically detect and report potholes. sent to a mobile application. Once a pothole is detected
The system consists of an accelerometer, a GPS nearby, an alert notification will be sent.
module, and a microcontroller. The accelerometer
measures the magnitude of acceleration and triggers In the paper titled 'Smart Pothole Detection Mapping
an alert when the acceleration exceeds a predefined System,' [2] the researchers introduced a novel system
threshold, indicating the presence of a pothole. The
that combines different sensors and machine learning
GPS module provides the location of the pothole,
algorithms to effectively detect potholes and create a
which is then sent to a remote server using Wi-Fi. By
following Google AppSheet. The system connects comprehensive map of their locations. To assess the
your Google Sheets containing date, time, longitude, system's performance, they conducted practical tests
and latitude data to an AppSheet app and presents the using a specially equipped vehicle fitted with the
data meaningfully within the app's interface. necessary sensors and algorithms. The vehicle was
Also, it involves analyzing image features and driven over roads with pre-identified potholes,
employing algorithms to detect pothole patterns. This allowing the researchers to validate and evaluate the
integration enables real-time monitoring of road system's accuracy and efficiency in real-world
conditions and accurate pothole detection. The conditions. According to the findings, the developed
system exhibited high accuracy in detecting potholes from the laser can be used to create an accurate model
and successfully generated a detailed map of their in digital form. Although the results are mostly
respective locations. The research paper highlights a accurate this approach is costly to implement.
cost-effective methodology that effectively identifies
potholes, making it a promising approach for pothole III. Hardware and Software
detection. Requirements
In the research paper titled 'Development of an
Intelligent System for Pothole and Hump Hardware requirements:
Identification on Roads,' [3] the authors propose a
comprehensive solution leveraging multiple sensors, 1. ESP8266 Wifi Module
such as a GPS receiver, accelerometer, and camera, to
gather data concerning the road surface. The data
collection process is outlined, along with the pre-
processing steps essential to preparing the collected
data for subsequent analysis.

[3] Furthermore, the study presents various machine

learning algorithms utilized for data analysis, which Fig 1: ESP8266 Wifi Module
encompass decision trees, support vector machines,
and artificial neural networks. The integration of these 2. The ESP8266 Wi-Fi Module is a self-
algorithms empowers the system to effectively contained System-on-Chip (SOC) that
identify and differentiate potholes and humps on the includes an integrated TCP/IP protocol stack,
road, making it a promising approach to enhance road providing any microcontroller with Wi-Fi
safety and maintenance. network connectivity. This versatile module
can act as a standalone application host or
The paper ‘Development and Analysis of Pothole offload Wi-Fi networking functions from
Detection and Alert based on NodeMCU’ [4] begins another application processor. Each ESP8266
module comes pre-loaded with firmware
by highlighting the problem of potholes on roads and
featuring an AT command set, making it
the dangers they pose to drivers. It then proceeds to
straightforward to integrate with an Arduino
describe the proposed system's architecture, which
device and harness Wi-Fi capabilities similar
consists of a NodeMCU microcontroller, a vibration to those offered by a Wi-Fi Shield.ADXL345
sensor, and an alert system. The vibration sensor is Accelerometer
attached to the bottom of a car and detects vibrations
caused by potholes. The NodeMCU microcontroller
processes these vibrations and sends an alert to the
driver through an LED light and a buzzer. The authors
present a detailed description of the system's
architecture, implementation, and advantages.

In addition, the paper ‘Analysis and improvements in

the current pothole detection system’[5] provides
various approaches that can be used for pothole 3. Fig 2: ADXL345 Accelerometer
detection and also explains about drawbacks of those
systems. A vision-based pothole detection system The ADXL345 accelerometer is capable of
employs image processing techniques to identify measuring both static acceleration due to
potholes. However, one of its limitations is its reliance gravity and dynamic acceleration resulting
on still images since obtaining images from moving from motion or shock. This compact sensor is
vehicles poses significant challenges. Another method available in a 14-lead LGA package, with
is 3D Reconstruction, in which the reflected pulse dimensions measuring just 3mm x 5mm x
 1mm. Due to its small size and versatile establish connections with a specified
capabilities, the ADXL345 finds applications internet IP address and port, which are
in various mobile device scenarios, including defined within the 'client.connect()' function.
mobile handsets, smartphones, gaming This secure client enables secure data
devices, pointing devices, and personal transmission over the internet, providing
navigation devices. Additionally, it is also enhanced protection for sensitive
utilized in applications like hard drive information.
protection, medical equipment, and industrial
instrumentation. 3. TinyGPS++ Library
The ‘TinyGPS.h’ library, also known as
4. GPS Module TinyGPS++, is a recently developed Arduino
library designed specifically for parsing
NMEA data streams originating from GPS
modules. Similar to its predecessor,
TinyGPS, this library offers concise and user-
friendly functions to extract essential
information like position, date, time, altitude,
speed, and course from GPS devices
commonly used by consumers. Its simplicity
and compactness make it an ideal choice for
handling GPS data with ease.
Fig 3: GPS Module

A GPS module receives location information

from satellites. It serves as an effective 4. Adafruit Library
method for determining precise locations The 'Adafruit_Sensor.h' library offers a user-
when used outdoors. The GPS module can friendly and device-independent interface for
access signals from multiple GPS satellites to seamless interaction between Adafruit IO and
accurately pinpoint a given location. Arduino. This library supports a wide range
of connectivity options, including Wi-Fi for
Software Requirements: ESP8266, ESP32, ESP32-S2, ESP32-S3,
ESP32-C3, Airlift, WINC1500, and WICED,
1. ESP8266Wifi library as well as Cellular for 32u4 FONA, and
The 'ESP8266WiFi.h' library facilitates the Ethernet for Ethernet Feather Wing. Its
functionality of the WiFi Module, enabling versatility allows users to switch between
the transmission of sensor-collected data to a different communication methods based on
server or client for storage. It utilizes TCP or their specific project requirements.
UDP communication protocols to establish
connections with the server or client. Devices 5. Adafruit ADXL345 Library
connecting to Wi-Fi networks are referred to The 'Adafruit_ADXL345_U.h' library is
as stations (STA), while the Wi-Fi access designed to measure both static acceleration
point (AP) serves as a hub for one or multiple due to gravity in tilt-sensing applications and
stations. The access point is further linked to dynamic acceleration resulting from motion
a wired network for data transfer and internet or shock. It offers several specialized sensing
connectivity. functions, including:

2. Wi-FiClient Library Activity and Inactivity Sensing: This feature

WifiClient The 'WiFiClientSecure.h' library detects motion presence or lack thereof by
allows the creation of a secure client that can
 comparing acceleration values on any axis The system architecture is illustrated in Figure 1. The
with user-defined thresholds. accelerometer and GPS module are connected to the
ESP8266 through the I2C interface. The ESP8266
Tap Sensing: The library can identify single processes the data from the accelerometer and GPS
module and sends the location of the pothole to a
and double taps in any direction.
Google spreadsheet using HTTPS.
Free-Fall Sensing: It can detect if the device is in free-
fall.
V. Methodology
These functions can be individually mapped to two
interrupt output pins, providing flexibility in triggering I. IOT Model working
specific actions. Moreover, the library incorporates an
integrated memory management system with a 32- The ESP8266 initializes the accelerometer and GPS
module and connects to the WiFi network.
level first-in, first-out (FIFO) buffer. This buffer The ESP8266 reads the acceleration data from the ac-
efficiently stores data, reducing the need for frequent accelerometer and calculates the magnitude of
interaction with the host processor and consequently acceleration.
lowering overall system power consumption. If the magnitude of the acceleration is greater than a
threshold value, the ESP8266 waits for a valid GPS
signal.
IV. System Architecture After acquiring a valid GPS signal, the ESP8266
The system consists of three main components: an retrieves the latitude and longitude data from the GPS
accelerometer, a GPS module, and a Node MCU. The module. Subsequently, the ESP8266 securely
accelerometer measures the magnitude of acceleration transmits this geographical information to a Google
in three axes, x, y, and z. The GPS module provides spreadsheet using HTTPS protocol. The Google
the locationof the system. The microcontroller acts spreadsheet can be accessed by authorized users to
as the intermediary between the accelerometer and view the location of the potholes detected by the
GPS module, processing the data from both sensors system.
and triggering an action when a potholeis detected.
The system utilizes the ADXL345 accelerometer,
renowned for its low-power consumption and
exceptional measurement accuracy. This 3-axis
accelerometer boasts high resolution with 13-bit
precision, allowing it to measure acceleration up to
±16g. Its capabilities make it a perfect fit for
accurately gauging acceleration magnitude, which is
crucial for pothole detection purposes.
The GPS module used in the system is TinyGPS++,
which is a compact and easy-to-use library for parsing
NMEAGPS data. This module provides the latitude
and longitudeof the system, which is used to identify
Fig 1: Code Snippets of the system
the location of the pothole.
The system incorporates the ESP8266
microcontroller, a cost-effective Wi-Fi microchip II. AppSheet
equipped with a complete TCP/IP stack and
microcontroller functionalities. This versatile To provide the date, time, longitude, and latitude data
microcontroller plays a crucial role in processing data from a Google Sheet to an AppSheet app we created a
obtained from the accelerometer and GPS module. Google Sheet and included columns for date, time,
When a pothole is detected, the ESP8266 triggers a longitude, and latitude and populated the sheet with
the relevant data. After enabling the Google Sheets
specific action, which involves sending the pothole's add-on Connected our Google Sheet to the AppSheet
location to a Google spreadsheet via HTTPS. This app. Then defined data types and formats in AppSheet
integration allows for efficient and reliable data as per our requirement for showing potholes data.
storage and analysis on the cloud platform. Afterward, the app interface displays and interacts
with the date, time, longitude, and latitude data. By r_image2`assigned the resized image r_image2 to the
following these steps, the system presents pothole data variable im. The imgray = cv2.cvtColor(im,
from the connected Google Sheet in a meaningful and cv2.COLOR_BGR2GRAY) converts the image im
user-friendly way within the AppSheet app. from BGR to grayscale using `cv2.cvtColor()` and
assigns it to the variable imgray. Later plt_show
(imgray) is used to display the grayscale image using
the `plt_show()` function defined earlier.
III. Image processing After getting the image in our conditions we set it
according to our pixels copy the image and store it in
In this project, we have used image processing to a different variable. Then we tend to draw contours to
figure out where is the pothole. This is done using the visualize the pothole using out =
Open CV library and we are visualizing the results cv2.drawContours(img2, contours2, -1, (0, 2050, 0),
using matplotlib.pyplot library. First, we have to 1). Its result is stored in the variable ‘out’. The
import the required libraries. They are: plt.title("drawContours Pothole Image") function sets
• cv2: OpenCV library for computer vision the title for the upcoming plot as "drawContours
tasks. Pothole Image". The ‘out’ image is then shown using
• numpy: Numerical computing library for array the plt.imshow() function where we pass the parameter
operations. out such as plt.imshow(out). plt.show() created the
• pygame: Library for multimedia applications, plots. Thus we can see an image with a plot and with
not used in the provided code. contours around the pothole.
• time: Library for time-related functions, not
used in the provided code.
• smtplib: The library facilitates sending emails
through the Simple Mail Transfer Protocol VI. Implementation
(SMTP) but was not utilized in the provided
code. In real life, the program could be implemented in a
• matplotlib. pyplot: Matplotlib library for device that is mounted on a vehicle or a bicycle. The
visualization. device would contain the necessary hardware,
including the ADXL345 accelerometer and GPS
At the start, we initialize the cv2 using module, and would be powered by a battery or by the
‘cv2.startWindowThread()’. This function is called to vehicle's electrical system. The program would run
initialize the thread for handling OpenCV's
windowing system. This is created as we have to view continuously, monitoring the acceleration and GPS
images. Then we used the ‘plt show(image, title="")’ data and detecting potholes when the acceleration
function. This is a custom function defined to display exceeds a predefined threshold. When a pothole is
an image using `matplotlib.pyplot`. It takes an image detected, the program would transmit the GPS data to
and an optional title as input. It checks the shape of the a Google Sheets spreadsheet using the HTTPS
image and converts it from BGR to RGB format if it protocol and the GAS_ID provided in the code. The
has three channels. It then sets the title and turns off spreadsheet would collect the GPS data, which could
the axis before displaying the image. The then be used to analyze road conditions and identify
image_resize(image, width=None, height=None,
areas that require maintenance.
inter=cv2.INTER_AREA) is another custom function
defined to resize an image. It takes an image, width,
height, and an interpolation method as input. It In addition to detecting potholes, the program could
calculates the aspect ratio based on the desired width also be modified to monitor other aspects of road
or height and resizes the image accordingly using the conditions, such as road surface temperature,
specified interpolation method. The resized image is humidity, and air pollution. This data could be used to
then returned. build a comprehensive database of road conditions and
to inform decisions about road maintenance and
To get the output with respect to a pic we have to infrastructure development.
import the pic. Loading and resizing an image is done
by r_image1 = cv2.imread('pic1.jpg'). the
cv2.imread() function is used to read the image. The Using HTTPS for data transmission in the proposed
r_image2 = image_resize(r_image1, width=275, pothole detection system is a prudent security
height=180) resizes the image ‘r_image1’ to a width measure; however, it introduces some privacy
of 275 pixels and a height of 180 pixels using the concerns that need careful consideration. While
`image_resize()` function. The resized image is HTTPS encrypts data during transmission, ensuring
assigned to the variable ‘r_image2’. confidentiality, it does not address potential privacy
issues related to the collection and storage of location
The first step of processing and displaying the image
is done by plt.title() function. Here is the data. The GPS module collects precise location
plt.title("Pothole Image") sets the title for the information, and in a larger-scale deployment, this
upcoming plot as "Pothole Image". im = could potentially lead to concerns about location
privacy. It is essential to implement strong access 15. Wait for the response from the Google Sheets
controls, data anonymization, and adherence to data server using the client.readStringUntil()
protection regulations to safeguard user privacy. function.
Additionally, informed consent and transparent data 16. Check whether the data was sent successfully
usage policies should be in place, ensuring that users or not by checking the response received
are aware of how their data is collected, stored, and from the server using the if statement.
used. Striking a balance between data collection for 17. Close the connection with the Google Sheets
pothole detection and user privacy is crucial to server using the client.stop() function.
maintaining public trust and compliance with privacy
regulations. Therefore, a comprehensive approach to VIII. Flowchart
privacy, encompassing both technical and policy
aspects, is necessary when implementing such
systems.

VII. Algorithm

1. Import required libraries: ESP8266WiFi,

WiFiClientSecure, TinyGPS++,
Adafruit_Sensor, and
Adafruit_ADXL345_U.
2. Initialize the ADXL345 accelerometer using
the accel.begin() function.
3. Print a message to indicate that pothole
detection has started.
4. Initialize the Wi-Fi connection with the SSID
and password using the WiFi.begin()
function.
5. Wait for the Wi-Fi connection to be
established by repeatedly checking the
connection status with the WiFi.status()
function.
6. Set the client object to use an insecure
connection.
7. In the loop() function, obtain acceleration
data from the ADXL345 accelerometer using
the accel.getEvent() function.
8. Calculate the magnitude of the acceleration
vector using the sqrt() function.
9. Check if the acceleration magnitude is
greater than the pothole detection threshold
using the if statement. IX. Results
10. If a pothole is detected, read the GPS
coordinates using the gps.encode() function.
This paper states how can potholes be detected using
11. Send the GPS coordinates to a Google Sheets
document using the sendData() function. the ESP8266 Wi-Fi module, GPS Module, and
12. Inside the sendData() function, establish a Accelerometer. Here the Wi-Fi module acts as an IOT
secure connection with the Google Sheets device where the data gets stored in the
server using the client.connect() function. microcontroller and displayed on the screen if a
13. Construct the URL for the Google Sheets pothole is detected the data is sent to an Excel sheet
document using the String() function. where the data of date, time, longitude, and latitude is
14. Send an HTTP GET request to the Google
Sheets server using the client.print() function. stored.
 Fig 6: Serial Monitor and Excel Output

This data can be sent to the government to analyze the

Fig 4: The connections with ESP8226 Wifi Module roads. The longitude and latitude values can be put on
Google Maps and we can find the exact location of the
When the device starts working it gives the values of pothole and we can repair it. This would lead to fewer
the Accelerometer and it measures the distance from road accidents and people can travel fast and safely.
the ground. If a pothole gets detected, it gives a text
saying “Pothole Detected” and stops giving the value Results of image processing:
from the accelerometer. Once the pothole is detected, if the image is captured,
we can see where are the potholes in the image. This
will lead to more accuracy than normal guessing of the
pothole. This is done by the OpenCV library which is
used to process and visualize the potholes. The
following image is an example of how the processed
image looks like.

Fig 5: Output of Pothole getting detected

Once the pothole gets detected, the date, time,

longitude, and latitude values that we have got from
the GPS Module gets stored at an excel sheet as shown
in Fig 6. A link of an excel sheet was added where the
data can be stored.

Fig 7: Output of Pothole getting detected

shows the bigger pothole clearly, and it even outlines

the difference in the roads where the pothole is and
where the pothole is not. By this image we can even
guess about the far-away potholes well, which
couldn’t be done by a normal image. By the contours, and municipalities. Our findings suggest that this
we can even see the severity of the pothole. approach can significantly enhance road infrastructure
quality, minimize disruptions to traffic flow, and
contribute to safer and more comfortable driving
Scalability of the project from the results
experiences. Further research and implementation of
generated: this technology can revolutionize road maintenance
practices and improve overall transportation
The successful deployment of the project on a larger infrastructure management.
scale hinge on several key considerations. First and
foremost is the scalability of the hardware The cost of this project came to around Rs. 1200.
components, ensuring they can be readily procured Multiple research papers and existing pothole detector
systems cost around Rs. 7000 to 10,000. This project
and integrated into a broader network or fleet of
requires a few improvements but the maximum cost
vehicles. Data transmission infrastructure must also be that we expect is around Rs. 4000 due to which this
robust enough to handle increased data volume system will be very helpful.
without bottlenecks, while data management and
storage systems must efficiently accommodate the In the future, the project aims to enhance pothole
amplified data generated in a large-scale deployment. detection accuracy by integrating advanced machine
Power supply and network infrastructure should be learning algorithms. Additionally, the team plans to
develop a dedicated mobile application that allows
addressed to ensure consistent and uninterrupted
users to report potholes in real time, thereby
operation. Software and analytics scalability is crucial contributing to more efficient road maintenance and
for real-time monitoring and data processing across a safety.
larger number of vehicles or sensors. The logistics of
installation and ongoing maintenance should be Another area of future research is the integration of
carefully planned to ensure smooth deployment. other sensors, such as cameras and LIDAR, to provide
Regulatory compliance must be navigated, and additional information about road conditions. This
could enable more precise pothole detection and
additional budgetary considerations should be made
classification, as well as the detection of other road
for the increased costs associated with scaling up. Pilot hazards such as cracks and bumps.
programs and testing should precede a full-scale
deployment to fine-tune the system and address any The ESP8266 microcontroller runs on the Arduino
unforeseen challenges. Collaboration with relevant programming environment, which is open-source and
authorities and stakeholders is key to the project's easy to use. It is also compatible with a wide range of
success on a larger scale. sensors and other peripherals, making it a popular
choice for IoT applications.

X. Conclusion The ESP8266 microcontroller runs on the Arduino

programming environment, which is open-source and
In In this paper, we have presented a system for easy to use. It is also compatible with a wide range of
automated pothole detection using an accelerometer sensors and other peripherals, making it a popular
and GPS module connected to an ESP8266 choice for IoT applications.
microcontroller. The system successfully detected and
recorded the location of potholes during field testing. References
This system can be used by local authorities to identify
areas with a high concentration of potholes and
prioritize repairs. [I]. Devekar, Nilesh, Swapnil Damodar, Prajakta
Shendkar, Wasim Mulani, and Vaibhav
By image analysis algorithms, we can accurately Narde. "Pothole detection system for
detect and locate potholes in real-time, providing monitoring road & traffic conditions using
valuable data for maintenance teams. The integration IoT." IJIRST-International Journal for
of NodeMCU, GPS, and accelerometer allows for
Innovative Research in Science &
precise geolocation and additional contextual
information about road conditions. This integrated Technology 5, no. 7 (2018): 34-37.
system enables proactive maintenance planning, [II]. Borgalli, Rohan. "Smart Pothole Detection
optimal resource allocation, and cost savings for cities and Mapping System." Journal of Ubiquitous
 Computing and Communication Conference on Advanced Logistics and
Technologies 2, no. 3 (2020): 136-144. Transport. IEEE, 2013.
[III]. Shivaleelavathi, B. G., Veeramma Yatnalli, [X]. Roadscanners oy. [Online]. Available:
S. Y. V. Chinmayi, and Spoorthi Thotad. http://www.roadscanners.com/
"Design and development of an intelligent [XI]. Real-time maps and traffic information based
system for pothole and hump identification on the wisdom of the crowd." [Online].
on roads." Int. J. Recent Technol. Eng 8, no. Available: http://world.waze.com/
3 (2019): 5294-5300. [XII]. "A Smart Pothole Detection System for
[IV]. Reddy, Etukala Jaswanth, Padhuri Navaneeth Urban Road Safety Management" by B. P.
Reddy, Govindula Maithreyi, M. Bharath Vijay Kumar, S. Rajeshwari, and B. V.
Chandra Balaji, Santanu Kumar Dash, and K. Kiranmayi.
Aruna Kumari. "Development and Analysis (https://ieeexplore.ieee.org/document/90237
of Pothole detection and Alert based on 17)
NodeMCU." In 2020 International [XIII]. "A Pothole Detection and Warning System
conference on emerging trends in Using Internet of Things (IoT)" by K. N.
information technology and engineering (ic- Nandini and K. S. Sathyanarayana.
ETITE), pp. 1-5. IEEE, 2020. (https://ieeexplore.ieee.org/document/86743
[V]. Srivastava, Sumit, Ayush Sharma, and Harsh 34)
Balot. "Analysis and improvements on [XIV]. "An IoT-based Pothole Detection and
current pothole detection techniques." Management System for Smart Cities" by R.
In 2018 International Conference on Smart Aravind, S. S. Sowmya, and S. Sridhar.
Computing and Electronic Enterprise (https://ieeexplore.ieee.org/document/89219
(ICSCEE), pp. 1-4. IEEE, 2018. 57) [X]. "Intelligent pothole detection system
[VI]. "Pothole detection and mapping system for based on accelerometer and machine learning
smart cities using deep learning and internet techniques," by K. R. Hossain, et al. in
of things," by S. Tripathi and S. Kumar in Measurement, Volume 173, 2021
International Journal of Electronics and [XV]. "A novel pothole detection method using
Communications, Volume 129, 2021. vehicle dynamics sensors," by N.
[VII]. "An Adaptive Neural Network-Based Jayasundara, et al. in International Journal of
Pothole Detection System for Intelligent Transportation Science and Technology,
Transportation Systems" by Mahesh A. Volume 10, Issue 2, 2021.
Laghate and Pravin A. Hangargekar (IEEE [XVI]. "Pothole detection and mapping system for
Transactions on Intelligent Transportation smart cities using deep learning and internet
Systems, 2019) of things," by S. Tripathi and S. Kumar in
[VIII]. "Pothole Detection System Using a Smart- International Journal of Electronics and
phone Sensor and Machine Learning Communications, Volume 129, 2021.
Techniques" by Yashna Sharma, et al. (IEEE [XVII]. "Pothole detection using accelerometer data
International Conference on Computing, and machine learning techniques," by M. S.
Power and Communication Technologies, Alam, et al. in Proceedings of the 2019
2020) International Conference on Machine
[IX]. “Real-Time Detection and Classification of Learning, Big Data, and Business
Potholes Using Low-Cost MEMS Intelligence, 2019.
Accelerometers" by D. H. Kim and K. W. [XVIII]. "Pothole detection using accelerometer data
Lee (IEEE Transactions on Intelligent and machine learning techniques," by M. S.
Transportation Systems, 2017) [12] Alam, et al. in Proceedings of the 2019
Elkosantini, Sabeur, and Saber Darmoul. International Conference on Machine
"Intelligent public transportation systems: A Learning, Big Data, and Business
review of architectures and enabling Intelligence, 2019.
technologies." 2013 International
[XIX]. An IoT-based Pothole Detection and
Management System for Smart Cities" by R.
Aravind, S. S. Sowmya, and S. Sridhar.
(https://ieeexplore.ieee.org/document/89219
57)
[XX]. Intelligent pothole detection system based on
accelerometer and machine learning
techniques," by K. R. Hossain, et al. in
Measurement, Volume 173, 2021