Analyzing structural characteristics of object category representations from their semantic-part distributions
This work addresses the need for interpretable analysis of object representations in computer vision, but it is incremental as it builds on existing sketch-based methods without introducing new paradigms.
The paper tackled the problem of understanding structural characteristics of object category representations by analyzing semantic-part distributions, and the result was a visualization method that provides intuitive insights into category-level structural trends using word clouds.
Studies from neuroscience show that part-mapping computations are employed by human visual system in the process of object recognition. In this work, we present an approach for analyzing semantic-part characteristics of object category representations. For our experiments, we use category-epitome, a recently proposed sketch-based spatial representation for objects. To enable part-importance analysis, we first obtain semantic-part annotations of hand-drawn sketches originally used to construct the corresponding epitomes. We then examine the extent to which the semantic-parts are present in the epitomes of a category and visualize the relative importance of parts as a word cloud. Finally, we show how such word cloud visualizations provide an intuitive understanding of category-level structural trends that exist in the category-epitome object representations.