Facial Key Points Recognition Data Science Competition The objective of the project is to determine the locations of key features on a face.
Objective is to develop a model that predicts positions of 15 key facial points - such as nose tip, centres and corners of eyes, mouth, lips, eyebrows, etc. - given images of faces, assuming the images are head-shots of the person and are in gray-scale.
For this project, R and Python was used. Python libraries include pandas, matplotlib, Theano and Lasagne;R libraries include ggplot, hmisc, IM.
The following is the list of the purpose of each script:
1.img_extract.py : Convert each image matrix into png image files.
2.preprocess_missing.py : Handle missing value.
3.ColMeansMethod.R : Implement the basic Column Means model.
4.PCA_script.R : Implement PCA on Training data: 7049 images.
5.MeanPatchScript.R :Implement the basic mean patch searching algorithm with -a) Image flipping -b) Contrast Stretching/Histogram Stretching -c) Histogram equalization -d) Gaussian Blurring
6.model1_NeuralNetwork.py :Python script for Neural Network.
7.model2_CNN.py :Python script for Convolutional Neural Network.
8.model3_CNN.py :Python script for Convolutional Neural Networks with Image flipping.