EyeOnArt

Computer Vision project for painting detection from videos.
The videos are captured in the art museum Galleria Estense, Modena.

Setup

Download the weights and cfg file from here.
Unzip the directory in the root dir of the project.

Setup the python environment :

$ pip install -r requirements.txt

Output

The outputs will be placed in the output dir.
There will be a view for each frame containing the bounding boxes for every detected painting. On top of every ROI you will find the title for the best match. Every function (that represents a step) of the pipeline is returning the requested output.

Usage Example

$ python EyeOnArt.py material/test.mp4

Help

usage: EyeOnArt.py [-h] [--skip N_FRAME] filename

-------------
|  EyeOnArt |
-------------
Painting Detection project.

               __________________________
             /|        Art Gallery       |
            / |  ____     ____     ____  |
           /  | |o   |   |  , |   | _  | |
          /   | |  O |   |.   |   |(@) | |
         /    | |_,k,|   |_,-,|   |\|p | |
        / /|  | |  h |   | ,; |   | |  | |
       / / |  | |_z__|   |____|   |____| |
      / /@;|  |  z z                     |
     /  |Y | z|_{)_______________________|
    /   | /  /z /H
   / /| |/  /z   Y
  / / |    / {)  d
 / / %|   /  /|
|  |&"|  /    Y
|  | /  /     d
|  |/  /
|     /
|    /
|   /
|  /
| /
|/

Credits :
Davide Casalini, Robert Covic & Stefano Rossi

positional arguments:
  filename        The filename to the source video you want to elaborate

optional arguments:
  -h, --help      show this help message and exit
  --skip N_FRAME  The number of frames to skip