The Effects of Lighting Conditions on the Accuracy and Security of Face ID

I chose this project because I use Face ID every day and started wondering how reliable and secure it really is in different situations. Since facial recognition is used in many devices, I wanted to test how well it works in real-world conditions.

The question I investigated was how lighting conditions affect both the accuracy and security of Face ID. I tested Face ID in bright light, dim light, and complete darkness, and with different facial obstructions such as masks, sunglasses, and regular glasses. I also tested security by seeing if another person or a photo of me could unlock my phone.

I found that Face ID works best in bright lighting when the face is fully visible. Masks and sunglasses reduced accuracy, especially when combined, while regular glasses had little to no effect. Face ID did not unlock for another person or a photo, showing that it is generally secure.

This project could benefit everyday smartphone users and companies developing facial recognition systems by helping them understand how environmental conditions affect performance and security. It can help improve the reliability of technology that many people depend on.

I started by researching how Face ID works using trusted websites like Apple’s support pages and technology education sites. This helped me understand that Face ID uses sensors to scan and recognize a person’s face.

For my experiment, I used my iPhone and tested Face ID under different conditions. I tested three lighting conditions: bright light, dim light, and complete darkness. I also tested different situations, including no obstruction, wearing a mask, sunglasses, regular glasses, and a combination of a mask and sunglasses.

For each condition, I performed 20 trials and recorded whether the phone unlocked successfully or not. I kept the phone at the same distance and angle each time to make sure the test was fair.

To test security, I had another person try to unlock my phone and also used a photo of myself. I recorded whether the phone unlocked or stayed locked.

All results were written down on data sheets and later turned into graphs to compare success rates. By keeping conditions consistent and repeating trials, I made sure my results were reliable.

The results of my project showed that both lighting conditions and facial obstructions affect how well Face ID works.

In bright lighting, Face ID worked very well when the face was fully visible, with a very high success rate. However, when a mask or sunglasses were worn, the success rate decreased. The lowest accuracy happened when both a mask and sunglasses were worn together, because most of the face was covered.

In dim lighting, Face ID still worked fairly well, but accuracy started to decrease in some conditions. Sunglasses had less impact than expected in dim light, but masks still reduced accuracy.

In complete darkness, Face ID had the lowest accuracy overall. Even though Face ID is designed to work in low light, the results showed that it was less reliable when there was no visible light.

Additional testing with regular glasses showed that they had little to no effect on accuracy. In most cases, Face ID still worked as expected because regular glasses do not block important facial features.

For the security part of the project, I tested whether another person or a photo of me could unlock my phone. In all trials, Face ID did not unlock for another person or a photo. This shows that the system is able to tell the difference between the correct user and someone else, and between a real face and a flat image.

Overall, the results show that Face ID works best when the face is clearly visible and lighting conditions are good. When important parts of the face are covered or when lighting is poor, accuracy decreases. However, the system remained secure and did not allow unauthorized access during testing.

From my results, I can conclude that both lighting conditions and facial obstructions affect the accuracy of Face ID. The system works best when the face is fully visible and in bright lighting. When important facial features are covered, such as with a mask or sunglasses, the accuracy decreases. The lowest performance occurred when both a mask and sunglasses were worn together, because most of the face was hidden. In darker lighting, Face ID was less reliable, even though it is designed to work in low light.

I also found that not all obstructions affect Face ID in the same way. Regular glasses had little to no impact on accuracy, while sunglasses had a greater effect because they block the eye area.

From the security tests, I learned that Face ID is generally secure. It did not unlock for another person or for a photo, which shows that it can tell the difference between the correct user and someone else.

Overall, I learned that Face ID depends on visible facial features and good conditions to work effectively. This shows that while it is reliable in everyday use, it can still be affected by real-world conditions.

If I continued this project, I would improve it by testing more people instead of only one user to see if the results stay the same. I would also test different phone models to compare how other facial recognition systems perform. Another improvement would be to use more controlled lighting levels to make the experiment more precise. For future research, I could test more advanced facial changes, such as different angles, distances, or prosthetic masks, to better understand how facial recognition works in real-world situations.

I would like to thank my teacher Dave Smith for supporting me throughout this project and for providing guidance on how to improve my experiment and presentation. His feedback helped me refine my research question and better understand my results. I also used online resources to learn how Face ID works, which helped me design my experiment and interpret my findings. Overall, this support and information helped make my project more complete, organized, and accurate.

Apple Inc. (2023). Use Face ID on your iPhone or iPad Pro. Apple Support. https://support.apple.com/en-ca/guide/iphone/iph6c6806d3/ios

Apple Inc. (2017, September 12). The future is here: iPhone X. Apple Newsroom. https://www.apple.com/newsroom/2017/09/the-future-is-here-iphone-x/

National Institute of Standards and Technology. (2022). Face recognition technology. https://www.nist.gov/programs-projects/face-recognition

HowStuffWorks. (n.d.). How facial recognition systems work. https://computer.howstuffworks.com/facial-recognition.htm

OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat