Code Artist

Obsessive
Compulsive
Orderer

The Joy of Finding

Trainspotters
Cat Ladies
Stamp Collectors
Jigsaw Puzzlers

Dopamine Junkies

The Brain's
Reward System

#bldigital

Beginnings

Analysis

127-dimensional Feature Vector

Global Features only

Mostly Scale and Rotation Invariant

Histograms

Haralick Texture Features

Proportions / Compression Ratios

The Tools

Step 1:
Download via Flickr API
and Analysis

Step 2:
Manual Classification
with File Explorer

Step 3a:
R & Rattle

Random Forests / SVMs

The downsides:

Training takes a long time

Upper limit for number of categories

Mixed success depending on category

The takeway:

Don't put all eggs in one basket.

Step 3b:
t-SNE Clustering

by Laurens van der Maaten

http://lvdmaaten.github.io/tsne

Diversion

The Internet Archive's Map of Book Subjects

Step 4:
The Limed Twig
or: Cherry Picking

Results

Decorative Elements

Recurring Themes:

"Lost at Sea"
"The Open Door"
"Dramas in Beds"

Cleaning

Using the Material