Prototypicality effects in global semantic description of objects

TL;DR

This paper presents a prototype-based semantic description method for objects that encodes their features using CNN-derived prototypes, enabling interpretable, discriminative, and typicality scoring aligned with human-like categorization.

Contribution

It introduces a novel semantic description model leveraging prototypicality effects, creating interpretable descriptors that reflect object typicality and distinctive features within categories.

Findings

Descriptors preserve CNN semantic information for classification

Distance metric effectively measures object typicality

Descriptors are semantically interpretable and simulate prototypical organization

Abstract

In this paper, we introduce a novel approach for semantic description of object features based on the prototypicality effects of the Prototype Theory. Our prototype-based description model encodes and stores the semantic meaning of an object, while describing its features using the semantic prototype computed by CNN-classifications models. Our method uses semantic prototypes to create discriminative descriptor signatures that describe an object highlighting its most distinctive features within the category. Our experiments show that: i) our descriptor preserves the semantic information used by the CNN-models in classification tasks; ii) our distance metric can be used as the object's typicality score; iii) our descriptor signatures are semantically interpretable and enables the simulation of the prototypical organization of objects within a category.