FACE-FAKE-NET USING ANN METHOD FOR IMAGE FACE ANTI

SPOOFING DETECTION

ABSTRACT:

Face Recognition (FR) systems are being used in a variety of applications, including road
crossings, banking, and mobile banking. The widespread use of FR systems has raised concerns
about the safety of face biometrics against spoofing attacks, which use the use of a photo or
video of a legitimate user's face to gain illegal access to the resources or activities. Despite the
development of several FAS or liveness detection methods (which determine whether a face is
live or spoofed at the time of acquisition), the problem remains unsolved due to the difficulty of
identifying discrimination and operationally reasonably priced spoof characteristics but also
approaches. Additionally, certain facial portions are frequently repeated or correlate to image
clutter, resulting in poor performance overall. This research proposes a face-anti-spoofing
neural network model that outperforms existing and live models.

In this project we present an anti-spoofing face method to solve the real-world scenario
that learns the target domain classifier based on samples used for training in a particular source
domain. Specifically, with the Artificial neural network (ANN), the Spatial/Channel-wise
Attention Modules were introduced. Two modules, namely the Spatial-wise Attention Module
and the Channel-wise Attention Module, were used at spatial and channel levels to improve
local features and ignore the irrelevant features. Extensive experiments on current collections
with benchmarks datasets verifies that the recommended solution will significantly benefit
from the two modules and better generalization capability by providing significantly improved
results in anti-spoofing. The proposed model outperforms the other state-of-the-art
methodologies. The proposed methodology is evaluated and verified on six benchmark datasets
i.e., NUAA Photograph Imposter Database, CASIA, Audience, The Images of Groups Dataset
(IOG), The Extended Cohn-Kanade Dataset CK+ and The Japanese Female Facial Expression
(JAFFE) Dataset
EXISTING SYSTEM:

 Using capsule networks to detect forged images and videos uses a method that uses a
capsule network to detect forged, manipulated images and videos in different scenarios,
like replay attack detection and computer-generated video detection.
 Exposing AI Created Fake Videos by Detecting Eye Blinking describes a new method to
expose fake face videos generated with deep neural network models. The method is
based on detection of eye blinking in the videos, which is a physiological signal that is
not well presented in the synthesized fake videos.

DISADVANTAGES:

 In their method, they have used random noise in the training phase which is not a good
option. Still the model performed beneficial in their dataset but may fail on real time
data due to noise in training.
 Their method only uses the lack of blinking as a clue for detection. However certain
other parameters must be considered for detection of the deep fake enchantment,
wrinkles on faces etc. Our method is proposed to consider all these parameters.

PROPOSED SYSTEM:

 We proposed The SpecDiff descriptor is made up of two types of reflections: a specular

reflection from the iris region, which has a specific brightness dependent on liveness,
and (ii) diffused reflections from the entire face region, which indicate the subject's 3D
structure. On an in-house database as well as four publicly available databases.
 Classifiers trained with SpecDiff descriptor beat existing flash-based PAD algorithms:

NUAA, Replay-Attack, Impersonating in the Wild, and OULU-NPU. Furthermore, when

compared to an end-to-end deep neural network classifier, the proposed technique

 achieves statistically significantly higher accuracy while having a six-fold faster execution

speed.

 The Spatial Aggregation of Bitmap Local Classifiers (SAPLC), which is made of an FCN

portion and an aggregation part, is proposed because of the investigation. The pixel-

level ternary labels, which contain the real foreground, faked foreground, and

undecided background, are predicted by the FCN component.

 The labels are then combined to produce an accurate image-level judgment.

Experiments on the CASIA-FASD, Replay-Attack, OULU-NPU and SiW datasets are also

carried out to statistically test the proposed SAPLC.

 The proposed SAPLC outperforms representative deep networks, such as two globally

supervised CNN models, one depth-based FCN, two FCNs with the class label, and two

FCNs with tetragonal labels, and achieves superior results close to some state-of-the-art

methods under various common protocols, according to the experiments.

ADVANTAGES:

 The output of the model is going to be whether the face image or face from live videos
is fake or a real along with the confidence of the model.
 Accuracy of the model is more compared, effective, and accurate to previous methods.
 Application can detect trained images in effective way compared to previous methods.

SOFTWARE HARDWARE REQUIREMENT:

HARDWARE REQUIREMENTS:
  System : i3 or above 2.4 GHz.
 Hard Disk : 500 GB.

 RAM : 8 GB and above.

SOFTWARE REQUIREMENTS:

 Operating system : Windows XP/7/10 and above

 Coding Language : Html, JavaScript

 Development Kit : Flask Framework

 Database : MySql

K. Priyanka

Name of the Student:

(Y21MTCS901)

Signature of the Guide:

Signature of the HOD: