Traffic sign detection and recognition based on Raspberry Pi
Author: Hasib Md. Abid Bin Farid, Md Tarequl Islam, Anas Ahmad, Md. Shahedul Islam , Marjuk Ahmed.
Abstruct: Traffic sign detection and recognition is an important feature in autonomous driving and reducing accidents by assisting the driver. This paper proposes an algorithm that will detect the traffic signs and identify them using different image processing methods, such as, threshold techniques, Gaussian blurring, masking, Contours parameters etc. For the evaluation purpose of this method a USB webcam and a raspberry pi have been used. The webcam is connected with the raspberry pi. The video is analyzed frame by frame and the algorithm is applied to each frame for the detection and classification purpose. Python programming language and openCV library has been used in this paper as software.
A Novel Approach Towards Railway Security System for Bangladesh Using Computer Vision
Author: Hasib Md Abid Bin Farid, Md. Fazla Rabbi, Tamjidun Nahar Nabila, Md. Abu Syeed, Sairin Nusrat Anam
Abstract: Rail accidents have turned into a very common problem in Bangladesh and incidences of accidents are recorded nearly every month. Some of the causes are common among guards, congested rail crossings, and the general ignorance of pedestrians. In addition to the injurious physical impact, these accidents result in huge losses in the economy as well as a large-scale disruption of traffic especially in high population centres like Dhaka. In order to tackle these issues, this paper discusses the use of Artificial Intelligence (AI) in improving safety on railway crossings. The proposed system identifies objects, including buses, rickshaws, cars and pedestrians by the use of image processing algorithms such as Gaussian filtering, thresholding, masking and contour analysis. Upon any object detection, an automated email notification is sent to the train operator by the system and it includes real-time updates on the occurrence at the crossing. This paper will design and implement the detection system and will discuss how it will help in mitigating rail accidents. The results show that the method has a potential in the future study and real world implementation in enhancing railway safety in Bangladesh.
IoT Based Smart Surveillance Security System Using Raspberry Pi
Author: Hasib Md Abid Bin Farid, Ridma Tabassum, Fazle Rabbi Riyad, Md. Rifatul Islam, Anindya Sanyal
Abstract: Border control is a highly significant aspect of national security, which is essential to prevent illegal translocation of people, animal species, and goods. The Republic of Bangladesh, a cost-effective country that is surrounded by India and Myanmar, is constantly engaged in the dispute related to illegal immigration, transnational crime, drug trafficking, and transference of fugitives. Therefore, the need to have effective border surveillance systems cannot be underscored. As a branch of the Ministry of Home Affairs, the Border Guards of Bangladesh (BGB) is charged with the responsibility of protecting the boundaries of the nation and must thus integrate advanced technologies to complement the operational capacity. Through this question, it is hypothesized to introduce the Internet of Things (IoT) approach to border surveillance systems, which will provide the BGB with the possibility to monitor in real-time. The desired system continuously records and transmits live video feeds which allows the operators to identify a target in the visual stream after which the equipment automatically enters into tracking the detected object in real time. Its algorithmic architecture is an implementation of the chromatic spectrum analysis and the use of advanced artificial intelligence protocols, based on the Open Computer Vision (OpenCV) library and Haar cascade classifiers to achieve object identification and tracking. The effectiveness of IoT-based object-tracking systems in the sphere of border security usage is supported by empirical data, which proves the accessibility of object-tracking systems and the possibility of significant influence.Finally, the study also aims at enlightening future technological advances and generating the pragmatism of implementing data-driven solutions based on border surveillance.
Hardware Implementation of Facial Recognition Using OpenCV
Author: Hasib Md Abid Bin Farid, Md.Rakib Hasan, Md. Maruf Billah, Shuvro Roy, Shuvo
Abstract: Assuring the safety of both residential and commercial sites has always been considered a priority matter, and the introduction of modern technology has a significant possibility to enhance existing protection systems. The current dissertation outlines a smart, real-time security and surveillance system that integrates a Raspberry Pi (RPi), a camera module, and an Android app and, therefore, provides a user-friendly and reliable surveillance framework. The RPi uses Haar Cascade classifier to recognize the faces of humans by extracting faces features of the images that are received. The system is trained to recognize individuals based on a repository of pre-stored images through adoption of a machine-learning approach with a basis on the Eigenfaces algorithm. The comparative analysis of captured frames with the data stored in the RPi SD card is made easier with the help of OpenCV, thus, allowing verification. The reference images can either be directly integrated with a computer interface or uploaded with the help of the adjunct Android application. In a case where the system comes across an unidentified face as an internet connection is made, the respective picture is automatically posted to a server that resides on the RPi, hence making it visible over the Android application. On the other hand, when there is a match with an existing database record, then nothing further is done. The Android application also provides the features of notification, image viewing, and storage, and switching the monitoring service off and on. By implementing such a system, users get prompt notifications about the activity in a designated area, therefore, communicating to them the possibility of an intrusion and storing image information that they can analyze at any given time they deem necessary. The given research testifies to the effectiveness of incorporating the methods of IoT and machine-learning to provide automated security solutions.