Projects per year
Abstract
Dense captioning (DC), which provides a comprehensive context understanding of images by describing all salient visual groundings in an image, facilitates multimodal understanding and learning. As an extension of image captioning, DC is developed to discover richer sets of visual contents and to generate captions of wider diversity and increased details. The state-of-the-art models of DC consist of three stages: (1) region proposals, (2) region classification, and (3) caption generation for each proposal. They are typically built upon the following ideas: (a) guiding the caption generation with image-level features as the context cues along with regional features and (b) refining locations of region proposals with caption information. In this work, we propose (a) a joint visual-textual criterion exploited by the region classifier that further improves both region detection and caption accuracy, and (b) a Geometryaware Relational Exemplar attention (GREatt) mechanism to relate region proposals. The former helps the model learn a region classifier by effectively exploiting both visual groundings and caption descriptions. Rather than treating each region proposal in isolation, the latter relates regions in complementary relations, i.e. contextually dependent, visually supported and geometry relations, to enrich context information in regional representations. We conduct an extensive set of experiments and demonstrate that our proposed model improves the state-of-the-art by at least +5.3% in terms of the mean average precision on the Visual Genome dataset.
Original language | English |
---|---|
Title of host publication | MULEA 2019 - 1st International Workshop on Multimodal Understanding and Learning for Embodied Applications, co-located with MM 2019 |
Publisher | ACM |
Pages | 3-11 |
Number of pages | 9 |
ISBN (Electronic) | 9781450369183 |
DOIs | |
Publication status | Published - 15 Oct 2019 |
MoE publication type | A4 Conference publication |
Event | International Workshop on Multimodal Understanding and Learning for Embodied Applications - Nice, France Duration: 25 Oct 2019 → 25 Oct 2019 Conference number: 1 |
Workshop
Workshop | International Workshop on Multimodal Understanding and Learning for Embodied Applications |
---|---|
Abbreviated title | MULEA |
Country/Territory | France |
City | Nice |
Period | 25/10/2019 → 25/10/2019 |
Keywords
- Attention
- Dense captioning
- Relationship modeling
Fingerprint
Dive into the research topics of 'Geometry-aware relational exemplar attention for dense captioning'. Together they form a unique fingerprint.Projects
- 2 Finished
-
MeMAD Laaksonen
Laaksonen, J. (Principal investigator)
01/01/2018 → 31/03/2021
Project: EU: Framework programmes funding
-
Deep neural networks in scene graph generation for perception of visual multimedia semantics
Laaksonen, J. (Principal investigator)
01/01/2018 → 31/12/2019
Project: Academy of Finland: Other research funding