GitHub - Ataraxia1001/Facial_Landmark_detection_Sticker_on_Face: CNN based facial landmark detection project

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 32 Commits
.ipynb_checkpoints		.ipynb_checkpoints
Stickers		Stickers
model		model
test_result		test_result
626176_0.txt		626176_0.txt
Brad_Pitt.png		Brad_Pitt.png
Readme.txt		Readme.txt
annotated_image.png		annotated_image.png
annotation.png		annotation.png
baby.jpg		baby.jpg
points on the face(LaPa).ipynb		points on the face(LaPa).ipynb
points on the face.ipynb		points on the face.ipynb
test_MobileNetV2.ipynb		test_MobileNetV2.ipynb
test_one_image.ipynb		test_one_image.ipynb
test_one_image_sticker_on.ipynb		test_one_image_sticker_on.ipynb
train_EfficientNetV2.ipynb		train_EfficientNetV2.ipynb
train_MobileNetV2.ipynb		train_MobileNetV2.ipynb

Repository files navigation

This is repository for facial landmark detection project.
It is detecting and recognizing the facial landmark by testing various convolutional neural networks.
Input data is face images and ouput of model is x, y coordinates values(points) which determine the facial landmarks 
like the positions of eyes, noes, mouth, and so on. The output has 106 differents points on the face.
I'm constantly attempt to use better model with more complexity, but the computing power currently makes impasse to run a model with 
high FLOPs.

I need to make multiple scripts to do this project.
Mainly, the codes are divided as follow.

1. plotting the points on the face.(train data)

2. training neural network(CNN model)

3. testing the model with new image(test data)

4. add sticker on the face image