Towards ontology driven learning of visual concept detectors

TL;DR

This paper presents an ontology-driven system that combines neural methods and active learning to enable real-time learning and detection of arbitrary visual concepts in videos, enhancing recall and semantic search capabilities.

Contribution

It introduces a novel framework integrating a large-scale visual ontology with neural and active learning techniques for dynamic concept detection.

Findings

Improved recall of concept detection through ontology guidance

Enables on-the-fly learning of new visual concepts

Provides semantic search in video libraries

Abstract

The maturity of deep learning techniques has led in recent years to a breakthrough in object recognition in visual media. While for some specific benchmarks, neural techniques seem to match if not outperform human judgement, challenges are still open for detecting arbitrary concepts in arbitrary videos. In this paper, we propose a system that combines neural techniques, a large scale visual concepts ontology, and an active learning loop, to provide on the fly model learning of arbitrary concepts. We give an overview of the system as a whole, and focus on the central role of the ontology for guiding and bootstrapping the learning of new concepts, improving the recall of concept detection, and, on the user end, providing semantic search on a library of annotated videos.