prem95/realtime-face-anti-spoofing

Realtime Face Anti Spoofing with Face Detector based on Deep Learning using Tensorflow/Keras and OpenCV

Hey there! I'm Prem Kumar

Check out my website 👉 Prem Kumar

What Do I Do?

• 🤖 I’m currently a Machine Learning Engineer in WISE AI building end to end eKYC pipeline for Face Anti Spoofing

• 📝 I regularly write articles on Medium

• 📄 Take a look at my LinkedIn profile LinkedIn

• 👋 I’m looking to collaborate on good projects that will improve my skills

Connect with me 🚀

Frameworks/Tools I use:

Support Me:

realtime-face-anti-spoofing's Issues

Model

Can you tell us how you trained the model?

Can this run on CPU-only computer?

Hi @Prem95
Can this run on CPU-only computer?
If so, how can I modify your code
thanks!

Can you share the anti spoofing model

Train model

Thanks for your great project. How can I train the model it will help me a lot.

In main.py file throwing an error "No module named Utility"

#21 from utility.video_utils import VideoUtils

On Github installation provides det_face folder. Cannot see a utility folder in face_det folder. Where should i get this folder?

How would i utilize it ?

Please provide me the steps to download this folder which utlized by this project.

No file named utility

Describe the bug
A clear and concise description of what the bug is.

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. iOS]
• Browser [e.g. chrome, safari]
• Version [e.g. 22]

Additional context
Add any other context about the problem here.

TypeError: 'NoneType' object is not subscriptable

File "main.py", line 61, in
frame_bgr = frame[..., ::-1]
TypeError: 'NoneType' object is not subscriptable

Using RBG or BGR?

https://github.com/Prem95/realtime-face-liveness-detector/blob/89a1109a18e6be227f02233ed6f36e15501fe106/main.py#L61

Hi,
The return value of cv2.VideoCapture(0) is based on BGR format by default. You should not convert it in line 61.

No module named 'FaceBoxes.utils.nms.cpu_nms'

Hi there I'm having this issue

Traceback (most recent call last):
  File "main.py", line 4, in <module>
    from utility.video_utils import VideoUtils
  File "C:\Users\hyper\OneDrive\Desktop\face-liveness-detector\utility\video_utils.py", line 9, in <module>
    from face_detector import FaceDetector
  File "C:\Users\hyper\OneDrive\Desktop\face-liveness-detector\face_detector.py", line 7, in <module>
    from FaceBoxes import FaceBoxes
  File "face_det\FaceBoxes\__init__.py", line 1, in <module>
    from .FaceBoxes import FaceBoxes
  File "face_det\FaceBoxes\FaceBoxes.py", line 10, in <module>
    from .utils.nms_wrapper import nms
  File "face_det\FaceBoxes\utils\nms_wrapper.py", line 10, in <module>
    from .nms.cpu_nms import cpu_nms, cpu_soft_nms
ModuleNotFoundError: No module named 'FaceBoxes.utils.nms.cpu_nms'

OS : Windows 11
Python Version 3.8.4

Runtime error found no NVIDIA driver on your system

I got this error when I try to run the main.py on macOS. Does the project require NVIDIA to run?

How to run it on windows

can you please tell me how i can run it on windows? since sh wont work on windows

There is no updated model

In your commit, you updated new model fas_94.h5

but there is no updated model. Only old model antispoofing.h5 existed.

No update?

Meaning of output_tensor[0]

Hello, congrats for great work.

I figure out the output_data has 2 positions.

np.round(output_data[:, 1][0], 3) .

In your example, you used the position 1 to calculate the predict score. What's the meaning of the position 0, in this model? Could it be the confidence about the prediction?

Access to the model

I tried to contact [email protected] to obtain access to the model file. However, i couldn't send mail to that adress. Can you help me to obtain access to the model file?

input shape of anti_spoofing.h5

If I want to use anti_spoofing.h5 separately, what image dimension should I pass to the model?

The model shows the following summary:

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #
=================================================================
 conv2d (Conv2D)             (None, 112, 112, 18)      504

 activation (Activation)     (None, 112, 112, 18)      0

 batch_normalization (BatchN  (None, 112, 112, 18)     72
 ormalization)

 conv2d_1 (Conv2D)           (None, 112, 112, 18)      2934

 activation_1 (Activation)   (None, 112, 112, 18)      0

 max_pooling2d (MaxPooling2D  (None, 56, 56, 18)       0
 )

 batch_normalization_1 (Batc  (None, 56, 56, 18)       72
 hNormalization)

 conv2d_2 (Conv2D)           (None, 56, 56, 36)        5868

 activation_2 (Activation)   (None, 56, 56, 36)        0

 batch_normalization_2 (Batc  (None, 56, 56, 36)       144
 hNormalization)

 conv2d_3 (Conv2D)           (None, 56, 56, 36)        11700

 activation_3 (Activation)   (None, 56, 56, 36)        0

 max_pooling2d_1 (MaxPooling  (None, 28, 28, 36)       0
 2D)

 batch_normalization_3 (Batc  (None, 28, 28, 36)       144
 hNormalization)

 flatten (Flatten)           (None, 28224)             0

 dense (Dense)               (None, 128)               3612800

 activation_4 (Activation)   (None, 128)               0

 dropout (Dropout)           (None, 128)               0

 dense_1 (Dense)             (None, 2)                 258

 activation_5 (Activation)   (None, 2)                 0

=================================================================
Total params: 3,634,496
Trainable params: 3,634,280
Non-trainable params: 216
_________________________________________________________________

What is the input shape?

Can you give me train script ?

Tks for your great project, can you update train script, it will help me a lot.

Missing utility.video_utils

Describe the bug
I'm trying to run the project but got an missing module error

➜  face-liveness-detector (main) ✔ python3 main.py

Traceback (most recent call last):
  File "main.py", line 4, in <module>
    from utility.video_utils import VideoUtils
ModuleNotFoundError: No module named 'utility'

To Reproduce
Steps to reproduce the behavior:

1. Clone project
2. Build face_det folder
3. Run python3 main.py
4. Got ModuleNotFoundError

Expected behavior
The project runs fine

Desktop (please complete the following information):

• OS: MacOS

onnx model?

hi @Prem95 ,
congrats for great work.

Is there onnx model available for antispoof.h5?

Thanks

How do I run your project successfully?

Hi @Prem95,
I tried to run your project, but it misses file: mb1_120x120.yml.
Can you provide it for testing?
Thank you!

Could you please release the code of model structure and how to train

Missing focal_loss package

Hi, please insert focal_loss==0.0.7 to requirements.txt file

$ sh build.sh ./build_cpu_nms.sh: line 2: python3: command not found ./build_sim3dr.sh: line 1: python3: command not found build.sh: line 10: gcc: command not found

Hi @prem, please provide some help to solve the issue.

$ sh build.sh
./build_cpu_nms.sh: line 2: python3: command not found
./build_sim3dr.sh: line 1: python3: command not found
build.sh: line 10: gcc: command not found