000			03493nam a22003257a 4500
003			OSt
005			20240812133813.0
008			240812s2024 |||||||| babm 00| 0 eng d
040			_aTUPM _beng _cTUPM _erda
050			_aBTH T 58.5 _bA44 2024
100			_aAleta, Chrystal Zhane H.
245			_aDevelopment of Fingerprint-Based Motorbike Ignition System with AI-Enabled Helmet Detection / _cChrystal Zhane H. Aleta, Ara Lou DM. Ancheta, Wynard I. Dela Rosa, Dave Matthew T. Diaz, Ma. Shaina R. Flocarencia.
264			_aManila : _bTechnological University of the Philippines, _c2024.
300			_ax, 132 pages : _billustrations ; _c29 cm. + _e 1 CD-ROM (4 3/4 in.)
336			_2rdacontent _atext
337			_2rdamedia _aunmediated
338			_2rdacarrier _avolume
500			_aThesis (Undergraduate)
502			_aCollege of Science -- _bBachelor of Science in Information Technology, _cTechnological University of the Philippines, _d2024.
504			_aIncludes bibliographical references.
520	3		_a"With the rising number of motorbikes being a primary mode of transportation, comes the need for enhancing the security of motorbikes and rider’s safety considering that motorbike theft and accidents have been evident for the past years. This research aimed to develop an enhanced Motorbike Ignition System advancing to keyless technology and integrating AI- Enabled Helmet Detection. The system allows identification of authorized users through a Fingerprint Sensor and improves rider’s safety through Artificial Intelligence, made possible by two controllers, specifically NodeMCU and Raspberry Pi 4B. The fingerprint sensor is attached to the NodeMCU enabling the motorbike’s electrical system while the Raspberry Pi is in charge of the Helmet Detection which starts the ignition system. Factoring in that the fingerprint sensor is affected if the finger is wet/oily and if with soil/rusty. For the helmet detection, lighting is vital for its functionality. Both conditions should be satisfied, else the ignition will not start. Adding a layer of security, three (3) consecutive failed attempts of fingerprint will activate an alarm. For the helmet detection feature, EfficientDet was used as algorithm. Through Raspbian OS, Google Colab, Tensorflow, Arduino IDE, Thonny IDE and Python, the software requirements were fulfilled. Datasets were prepared in helmet detection training ensuring dependability. Model Performance Evaluation showed that the system’s Mean Average Precision (mAP) is 94.37% while the Average Precision (AP) is 0.9436502, indicating the system’s high precision. The system was also subjected to different test cases to determine needed improvements. The system was then evaluated by IT professionals, electricians, and motorists using the TUP evaluation instrument. Majority of the respondents highly accepted the system in terms of Functionality, Aesthetics, Workability, Durability, Economy and Safety. With a grand weighted mean of 3.59 interpreted as “Highly Acceptable”, the system was able to accomplish its objective, improving the current security of motorbikes and safety of the riders." -- Author's Abstract
650			_aMotorcycles _vMotors
650			_aFingerprints _vIdentification
700			_aAncheta, Ara Lou DM.
700			_aDela Rosa, Wynard I.
700			_aDiaz, Dave Matthew T.
700			_aFlocarencia, Ma. Shaina R.
942			_2lcc _cBTH COS _n0
999			_c28834 _d28834