Behind the Scenes
All the data of the Buddha sculptures and major images are requested from the Met's API. The facial data processing is conducted by Open-CV and Scikit-Image.
Scikit-image's pixel process functions are used to enhance the images.
Open-CV is used to extract facial information from Buddhas' faces. 68 landmarks are extracted from each face. 211 out of 416 faces are successfully extracted. They are clustered in 12 different culture regions, covering the main areas that Buddhism is commonly practiced.
Three different visualizations: face dots, lines & shapes, and face grid, are utilized to demonstrate and compare the facial features of Buddha sculptures.
Steps:
- Design & Sketch
- Request Buddha sculptures' information and major images from the Met's API.
- Clean the data. Find better images if the major images from the request are not frontal shoot.
- Using Open-CV and Scikit-Image to process images.
- Using Open-CV to extract facial information from the Buddha faces.
- Front-end development with D3.js.
Partner: The Metropolitan Museum of Art
Data Source: The Met Collection API
Date: 2019