Brown Dog R Package (BD-R):

Installation

• Download R from anyone of the mirror site: https://cran.r-project.org/mirrors.html
• Install R following the instructions provided by the website: https://cran.r-project.org/doc/manuals/r-release/R-admin.html

library(devtools)
install_github("NCSABrownDog/bd.r")
library(BrownDog)

• A set of example files for test can be found here : http://browndog.ncsa.illinois.edu/examples/caltech101/

• Index method creates .index.tsv file with file name and correspondingg tags and content descriptors obtained from Brown Dog service. An example .index.tsv file content is shown below:

  airplane.jpg [[0.0721,0.0508,0.777,0.1001,0,0,0,0],[0.1502,0.2493,0.5582,0.0424,0,0,0,0],[0.137,0.0398,0.7637,0.0594,0,0,0,0]]

  brontosaurus.jpg

  cougar.jpg [[0.2571,0.5403,0.1928,0.0098,0,0,0,0],[0.4692,0.5001,0.0308,0,0,0,0,0],[0.3381,0.59,0.0713,0.0007,0,0,0,0]]

  dollar_bill.jpg ["Human Face Automatically Detected"] ["Person Automatically Detected"] [[0.2035,0.2141,0.5724,0.0101,0,0,0,0],[0.2509,0.2266,0.5225,0,0,0,0,0],[0.1939,0.2189,0.5707,0.0165,0,0,0,0]]

  person.jpg ["Mid Close Up Automatically Detected"] ["Human Face Automatically Detected"] ["Person Automatically Detected"] ["Human Eyes Automatically Detected"] ["Human Profile Automatically Detected"] [[0.1969,0.393,0.4079,0.0022,0,0,0,0],[0.1983,0.6102,0.1912,0.0004,0,0,0,0],[0.177,0.5175,0.3048,0.0007,0,0,0,0]]

• Find method allows to find similar images in a folder for a given query image using the .index.tsv file. An example output looks like:

  [1] "Distance of each file to the query file" containing 5 key-value pair(s). airplane.jpg : 0.2321195 brontosaurus.jpg : 1.797693e+308 cougar.jpg : 0.5819163 dollar_bill.jpg : 0 person.jpg : 0