Details
| Original language | English |
|---|---|
| Title of host publication | Pattern Recognition |
| Subtitle of host publication | 41st DAGM German Conference, DAGM GCPR 2019, Proceedings |
| Place of Publication | Cham |
| Publisher | Springer Nature |
| Pages | 331-344 |
| Number of pages | 14 |
| ISBN (print) | 9783030336752 |
| Publication status | Published - 25 Oct 2019 |
| Event | 41st DAGM German Conference on Pattern Recognition, DAGM GCPR 2019 - Dortmund, Germany Duration: 10 Sept 2019 → 13 Sept 2019 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 11824 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (electronic) | 1611-3349 |
Abstract
Visual relationship detection targets on predicting categories of predicates and object pairs, and also locating the object pairs. Recognizing the relationships between individual objects is important for describing visual scenes in static images. In this paper, we propose a novel end-to-end framework on the visual relationship detection task. First, we design a spatial attention model for specializing predicate features. Compared to a normal ROI-pooling layer, this structure significantly improves Predicate Classification performance. Second, for extracting relative spatial configuration, we propose to map simple geometric representations to a high dimension, which boosts relationship detection accuracy. Third, we implement a feature embedding model with a bi-directional RNN which considers subject, predicate and object as a time sequence. We evaluate our method on three tasks. The experiments demonstrate that our method achieves competitive results compared to state-of-the-art methods.
ASJC Scopus subject areas
- Mathematics(all)
- Theoretical Computer Science
- Computer Science(all)
- General Computer Science
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Pattern Recognition: 41st DAGM German Conference, DAGM GCPR 2019, Proceedings. Cham: Springer Nature, 2019. p. 331-344 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11824 LNCS).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Exploiting Attention for Visual Relationship Detection
AU - Hu, Tongxin
AU - Liao, Wentong
AU - Yang, Michael Ying
AU - Rosenhahn, Bodo
N1 - Funding Information: We thank CERN for the very successful operation of the LHC, as well as the support staff from our institutions without whom ATLAS could not be operated efficiently. We acknowledge the support of ANPCyT, Argentina; YerPhI, Armenia; ARC, Australia; BMWFW and FWF, Austria; ANAS, Azerbaijan; SSTC, Belarus; CNPq and FAPESP, Brazil; NSERC, NRC and CFI, Canada; CERN; CONICYT, Chile; CAS, MOST and NSFC, China; COLCIENCIAS, Colombia; MSMTCR, MPOCR and VSC CR, Czech Republic; DNRF, DNSRC and Lundbeck Foundation, Denmark; IN2P3-CNRS, CEADSM/ IRFU, France; GNSF, Georgia; BMBF, HGF, and MPG, Germany; GSRT, Greece; RGC, Hong Kong SAR, China; ISF, I-CORE and Benoziyo Center, Israel; INFN, Italy; MEXT and JSPS, Japan; CNRST, Morocco; FOM and NWO, Netherlands; RCN, Norway; MNiSW and NCN, Poland; FCT, Portugal; MNE/IFA, Romania; MES of Russia and NRC KI, Russian Federation; JINR; MESTD, Serbia; MSSR, Slovakia; ARRS and MIZ?, Slovenia; DST/NRF, South Africa; MINECO, Spain; SRC and Wallenberg Foundation, Sweden; SERI, SNSF and Cantons of Bern and Geneva, Switzerland; MOST, Taiwan; TAEK, Turkey; STFC, United Kingdom; DOE and NSF, United States of America. In addition, individual groups and members have received support from BCKDF, the Canada Council, CANARIE, CRC, Compute Canada, FQRNT, and the Ontario Innovation Trust, Canada; EPLANET, ERC, FP7, Horizon 2020 and Marie Sk?odowska-Curie Actions, European Union; Investissements d?Avenir Labex and Idex, ANR, Region Auvergne and Fondation Partager le Savoir, France; DFG and AvH Foundation, Germany; Herakleitos, Thales and Aristeia programmes co-financed by EU-ESF and the Greek NSRF; BSF, GIF and Minerva, Israel; BRF, Norway; the Royal Society and Leverhulme Trust, United Kingdom. The crucial computing support from all WLCG partners is acknowledged gratefully, in particular from CERN and theATLAS Tier- 1 facilities at TRIUMF (Canada), NDGF (Denmark, Norway, Sweden), CC-IN2P3 (France), KIT/GridKA (Germany), INFN-CNAF (Italy), NL-T1 (Netherlands), PIC (Spain), ASGC (Taiwan), RAL (UK) and BNL (USA) and in the Tier-2 facilities worldwide.
PY - 2019/10/25
Y1 - 2019/10/25
N2 - Visual relationship detection targets on predicting categories of predicates and object pairs, and also locating the object pairs. Recognizing the relationships between individual objects is important for describing visual scenes in static images. In this paper, we propose a novel end-to-end framework on the visual relationship detection task. First, we design a spatial attention model for specializing predicate features. Compared to a normal ROI-pooling layer, this structure significantly improves Predicate Classification performance. Second, for extracting relative spatial configuration, we propose to map simple geometric representations to a high dimension, which boosts relationship detection accuracy. Third, we implement a feature embedding model with a bi-directional RNN which considers subject, predicate and object as a time sequence. We evaluate our method on three tasks. The experiments demonstrate that our method achieves competitive results compared to state-of-the-art methods.
AB - Visual relationship detection targets on predicting categories of predicates and object pairs, and also locating the object pairs. Recognizing the relationships between individual objects is important for describing visual scenes in static images. In this paper, we propose a novel end-to-end framework on the visual relationship detection task. First, we design a spatial attention model for specializing predicate features. Compared to a normal ROI-pooling layer, this structure significantly improves Predicate Classification performance. Second, for extracting relative spatial configuration, we propose to map simple geometric representations to a high dimension, which boosts relationship detection accuracy. Third, we implement a feature embedding model with a bi-directional RNN which considers subject, predicate and object as a time sequence. We evaluate our method on three tasks. The experiments demonstrate that our method achieves competitive results compared to state-of-the-art methods.
UR - http://www.scopus.com/inward/record.url?scp=85076175014&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-33676-9_23
DO - 10.1007/978-3-030-33676-9_23
M3 - Conference contribution
AN - SCOPUS:85076175014
SN - 9783030336752
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 331
EP - 344
BT - Pattern Recognition
PB - Springer Nature
CY - Cham
T2 - 41st DAGM German Conference on Pattern Recognition, DAGM GCPR 2019
Y2 - 10 September 2019 through 13 September 2019
ER -