NUST Institutional Repository

HYBRID LIE DETECTOR: USING FACIAL MICRO-EXPRESSIONS AND PHYSIOLOGICAL DATA

Show simple item record

dc.contributor.author MUHAMMAD OSAMA , ZAINAB ZAHID
dc.date.accessioned 2025-02-13T07:31:45Z
dc.date.available 2025-02-13T07:31:45Z
dc.date.issued 2024
dc.identifier.uri http://10.250.8.41:8080/xmlui/handle/123456789/49849
dc.description Advisor: Prof. Dr. Imran Usman en_US
dc.description.abstract Lie detection is a way to gather truth-based, useful information from people under precarious circumstances. Often, the variations in human traits upon exhibiting deceit are ignored by humans, resulting in an unsavoury situation involving the authorities chasing the same goons whom they had already captured but let them go due to inconclusive evidence. Pakistan’s harrowing crime rate, and a lack of an efficient and effective system to oust the suspects during interrogation without jeopardizing their human rights, have led us to develop a smart solution to this problem once and for all. Our project delivers a hybrid model, a combination of intrusive and non-intrusive approaches of lie detection. Monitoring of the suspect takes place intrusively via pulse sensor. Additionally, non-intrusive monitoring would be taking place side-by-side via a camera, input of which will be preprocessed, and various automation methodologies will be applied on it, after which comparisons are made against datasets available for micro expressions. This results in our hybrid approach which, combines the two steps into one which leads to an improved decision making in controlled environment. Limitation of this project are currently due to the lack of dataset of people with deformities on their face, which makes it extremely difficult to calculate their micro expressions with automated methodologies. This project also lacked in terms of the hardware since we were not able to get a RaspberryPi Camera Module as well as a RaspberryPi 4, which made us resort to using our webcam to record micro-expressions and using an Arduino Uno as the micro-controller. The direct consequence of this on our project was that it resulted in a very jittery camera-feed, and we faced difficulty in compiling our fused dataset for our Lie Detection model to be trained on. It also caused a lot of “incorrect” pulse sensor input due to the use of an Arduino. However, the solution which we present can be transformed into a collaborative project with the intelligence agencies of Pakistan to elevate the living standards of the common folk, by aiding them through a renewed sense of security and safety. en_US
dc.language.iso en en_US
dc.title HYBRID LIE DETECTOR: USING FACIAL MICRO-EXPRESSIONS AND PHYSIOLOGICAL DATA en_US
dc.type Thesis en_US


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account