Bangkit 2021 Capstone Team : CAP0166

Hi. this is our repository for our project. Our team consist of 2 people from each Bangkit Path which is consist of Machine Learning, Android, and Cloud Computing.

All of the project management will take place on Github using the project board.

Our Team Member

Nama	Bangkit-ID	Path
Muhammad Zein I. F.	M2582405	Machine Learning
Yusril Hasanuddin	M2582412	Machine Learning
Salsha Farahdiba	C2472321	Cloud Computing
Fachry	C2582415	Cloud Computing
Salsabila Khairunnisa	A0050361	Android
Antoni Kurniawan	A0050366	Android

What Are We Doing In This Project

We make automatic students presence by using both Face detection and Face recognition to identify a person based on their faces

Tech Stack

Local Deployment Machine Learning Jupyter Notebook

Make sure you intalled this dependency first on your local machine. You can use virtual-env or conda virtual env for making things easier

scikit-image==0.18.1
facenet-pytorch==2.5.2
matplotlib==3.4.2
validators==0.18.2
opencv-python==4.5.2.52
scikit-learn==0.24.2
mtcnn==0.1.0
tensorflow==2.4.1

For using our machine learning example first you need to clone our project or fork our project by using this line

git clone https://github.com/mozarik/bangkit-capstone-CAP0166.git bangkit-project

after that there is a folder called bangkit-project. The next step is Generating Your Dataset

There is 2 way to generate your training data.

• By deepfaking your photo by using generate_data_deepfake.ipynb or you can go to this notebook Generate Deepfake Data
• Or you can use your selfie photo make sur to use 1:1 resolution for better accuracy.

After you done making the dataset go first you need to make a directory structure like this. The name filename does not matter as long you put the Name of the person as the directory.

For training the data you will need a image of a person face but in that image there's only one picture

├───image
│   ├───Brad_Pitt
│   │       bradd_1.jpg
│   │       bradd_2.jpg
│   │       bradd_3.jpg
│   │       bradd_4.jpg
│   │       bradd_5.jpg
│   │
│   ├───Yusril
│   │       yusril_1.jpg
│   │       yusril_2.jpg
│   │       yusril_3.jpg
│   │       yusril_4.jpg
│   │       yusril_5.jpg
│   │       yusril_6.jpg
│   │       yusril_7.jpg
│   │       yusril_8.jpg

cd to /ml-projectAnd then run. python .\train.py <your-image_dataset-directory> model\facenet_keras_weights.h5

after this you will see a encodings.pkl

You can use encodings.pkl as your embedding file.

For example see playground_example_zein.ipynb to run the inference.

Project Demo

See this video on youtube Video