Anti Face Spoofing Detection
Cybercrime is on the rise in our increasingly digital world. Many organization are now exploring biometric face recognition as a viable security solution offered by machine learning engineers. But in facial recognition, there are some problems. User photos can easily be found through social networks
2025-06-28 16:30:15 - Adil Khan
Anti Face Spoofing Detection
Project Area of Specialization Artificial IntelligenceProject SummaryCybercrime is on the rise in our increasingly digital world. Many organization are now exploring biometric face recognition as a viable security solution offered by machine learning engineers. But in facial recognition, there are some problems. User photos can easily be found through social networks and used to spoof facial recognition software. Let us say using paper photographs, screenshots, or facial reconstruction. So the face anti-spoofing is a technique that could prevent a face-spoofing attack. For example, violator might use a photo of the legal user to "deceive" the face recognition system. Therefore, it is important to use the face anti-spoofing technique to enhance the security of the system. We will use the dataset of CASIA- FASD.
• To detect the face spoofing
• To compare real & spoofed image
• To detect false facial verification using a photo, video, mask or different substitute of the authorized person.
Anti-face spoofing detection is a Realtime 2D image face recognition detection, which detects whether or not image presented is spoofed or real image. For this we will pass the input through CNN algorithm then after the process, we get output weather image is spoofed or not. Image is pass through a pipeline of CNN in which data will be compared with several algorithms based on their weight, bias, Gaussian filter & many more. The dataset which we will use for this purpose is CASIA-FASD, we will train dataset after cleaning data in such a way that it will predict output with maximum accuracy. After train the model we will use the pickle file of the model for our project so there will be less computation.
Benefits of the ProjectBenefits of the project are important for the organization to have face anti-spoofing systems in place to protect sensitive data, reduce theft, and reduce fraud. These systems improve existing facial recognition solutions by enhancing their ability to detect fraud. We can addon this product with other security measures to enhance security. This product can make facial biometric verification reliable after fingerprint. The product can be used in gateways so a person does not have to put a thumb or any other thing on a scanner for verification purpose. The camera will detect it without making any move from the individual.
Technical Details of Final DeliverableWe can install our product on biometric attendance systems so that is will alert us weather user marking attendance is authentic or not. For this purpose, a High deficiency camera with take input of individual face features & pass it through the pipeline of CNN which process it from all aspects to detect it where it spoofing attack or not by detecting face liveness, the contextual information of an image by investigating its surrounding, texture analysis that picture have regular patterns or not & so on features. Dataset we are using for this precise task is CASIA-FASD which would be trained in a way that it predicts that image is spoofed or not. If the image is spoofed then LCD will display an alert message that it’s a spoofing attack.
We will use raspberry Pi 4 for computation & processing, a camera module with be attached to Pi 4 with male to male wires. Camera module with taking input images then store it in class 10 SD card. Input will be pass through the pipeline of CNN which process it from all aspects to detect it where it spoofing attack or not by detecting face liveness after that Pi 4 with send output result on the LCD which would be attached to it with HDMI cable. A raspberry Pi 4 case will be used to protect it from dust, dirt & direct sunlight. For the power supply, we will use a type-C micro USB adapter to supply power to the Pi module. To make sure stable working & making it portable a Li-ion type-C Pi 4 battery will be used, an 8 channel relay will be used to protect pieces of equipment from high voltage supply like LCD.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 73250 | |||
| Raspberry Pi 4 | Equipment | 1 | 14000 | 14000 |
| Camera | Equipment | 2 | 7000 | 14000 |
| Male to male wires | Equipment | 10 | 400 | 4000 |
| Battery 12v | Equipment | 1 | 6000 | 6000 |
| LCD | Equipment | 1 | 7000 | 7000 |
| Button | Equipment | 1 | 150 | 150 |
| Pi 4 Cover Case | Equipment | 1 | 3500 | 3500 |
| Power Supply | Miscellaneous | 1 | 5000 | 5000 |
| HDMI Cable | Miscellaneous | 1 | 2500 | 2500 |
| Pi 4 relay 8 channel module | Equipment | 1 | 1800 | 1800 |
| Pi 4 bettery type-C | Equipment | 1 | 11000 | 11000 |
| 32 Gb class 10 SD card | Equipment | 1 | 4300 | 4300 |