Large Scale Visual Art Workshop

Large Scale Visual Recognition

Visual recognition remains one of the grand goals of artificial intelligence research. One major challenge is endowing machines with human ability to recognize tens of thousands of categories. Moving beyond previous work that is mostly focused on hundreds of categories, we make progress toward human scale visual recognition. Specifically, our contributions are as follows: First, we have constru...

Workshop Report: Large-Scale Parallel KDD Systems

Large-Scale Analysis of Art Proportions

While literature often tries to impute mathematical constants into art, this large-scale study (11 databases of paintings and photos, around 200.000 items) shows a different truth. The analysis, consisting of the width/height propor‐ tions, shows a value of rarely if ever one (square) and with majority of images having a proportion larger than one, but less than e.g. the golden ratio. Further‐ ...

Toward an Objective Value Scale of Creativity in Visual Art

ion 26 – 50 25% or 1/4 Imagination 51 – 75 25% or 1/4 Conceptualization 76 – 100 25% or 1/4 In the above example, each element acquires an equal range of values with weighted differences between elements according to the creative value of each element. If a work of visual art lacks substantial conceptualization, resulting value ranges reflect a substantial increase in distributed weight to acco...

Large-Scale Visual Odometry for Rough Terrain

Motion estimation from stereo imagery, sometimes called visual odometry, is a well-known process. However, it is difficult to achieve good performance using standard techniques. We present the results of several years of work on an integrated system to localize a mobile robot in rough outdoor terrain using visual odometry, with an increasing degree of precision. We discuss issues that are impor...