01/13~01/14 UBC GirlCode Hackathon project: Capsule Closet
Input an image, produce the season color that suits the skin tone the best.
part1-train model for classification
files:
- step1-processdata.ipynb --> process the csv file, turn rgb into hsv, manually label 123 dataset
- step2-training.ipynb --> use svm to train the data, map the image to its corresponding seasonal color type
output:
- datall.csv: the file that contains the processed csv file
- model.pickle: the trained model
data source:
- skin tone dataset: https://github.com/the-pudding/data/blob/master/vogue/models.csv
part2-implement the model
files:
- wholeline.ipynb --> Input an image, output its image with color palette on top.
framework:
additional part: the color for four classes
files:
- autumn.jpg: the color palette picture for autumn
- spring.jpg: the color palette picture for spring
- summer.jpg: the color palette picture for summer
- winter.jpg: the color palette picture for winter
- d-getrgb.ipynb: function that converts the above .jpg into a list of (r,g,b) values
output:
- mapseason.pickle: the dictionary that maps season to its corresponding list of 3 (r,g,b) colors