Bertrand Delezoide
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Bertrand Delezoide has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Alexandru Lucian Ginsca
CEA, LIST, Gif-sur-Yvette, France
2 shared publications - Françoise Prêteux
CERMICS, Ecole des Ponts, Champs-sur-Marne, France
2 shared publications - FrançoIs Plesse
CEA, LIST, Gif-sur-Yvette, France
1 shared publication François Plesse
1 shared publication
Latest Publications
- 4 August 2020
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (15) , 22465-22486
Modelling relations with prototypes for visual relation detection
- 4 August 2020
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 80 (15) , 22465-22486
Modelling relations with prototypes for visual relation detection
- Open Access
Visual Relationship Detection Based on Guided Proposals and Semantic Knowledge Distillation
A thorough comprehension of image content demands a complex grasp of the interactions that may occur in the natural world. One of the key issues is to describe the visual relationships between objects. When dealing with real world data, capturing these very diverse interactions is a difficult problem. It can be alleviated by incorporating common sense in a network. For this, we propose a framework that makes use of semantic knowledge and estimates the relevance of object pairs during both training and test phases. Extracted from precomputed models and training annotations, this information is distilled into the neural network dedicated to this task. Using this approach, we observe a significant improvement on all classes of Visual Genome, a challenging visual relationship dataset. A 68.5% relative gain on the recall at 100 is directly related to the relevance estimate and a 32.7% gain to the knowledge distillation.
Open AccessVisual Relationship Detection Based on Guided Proposals and Semantic Knowledge Distillation
A thorough comprehension of image content demands a complex grasp of the interactions that may occur in the natural world. One of the key issues is to describe the visual relationships between objects. When dealing with real world data, capturing these very diverse interactions is a difficult problem. It can be alleviated by incorporating common sense in a network. For this, we propose a framework that makes use of semantic knowledge and estimates the relevance of object pairs during both training and test phases. Extracted from precomputed models and training annotations, this information is distilled into the neural network dedicated to this task. Using this approach, we observe a significant improvement on all classes of Visual Genome, a challenging visual relationship dataset. A 68.5% relative gain on the recall at 100 is directly related to the relevance estimate and a 32.7% gain to the knowledge distillation.