AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Graphical Contrastive Losses for Scene Graph Parsing

Ji Zhang, Kevin J. Shih, Ahmed Elgammal, Andrew Tao, Bryan Catanzaro

CVPR 2019 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 268 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add all three graphical contrastive losses to predicate cross-entropy training (ablation)RelDN · OpenImages miniwmAP_rel: 34.63 -> 35.54Improved
add all three graphical contrastive losses to predicate cross-entropy training (ablation)RelDN · OpenImages miniAP_rel plays: 36.04 -> 41.04 (+5.0)Improved
add all three graphical contrastive losses (ablation)RelDN · Visual Genome SGDETRecall@50: 28.1 -> 28.3Improved
add all three graphical contrastive losses (ablation)RelDN with COCO initialization · VRD relationship detection, free kRecall@50: 26.67 -> 28.15Improved
use class-agnostic contrastive loss L1 (ablation)RelDN · OpenImages miniwmAP_rel: 34.63 -> 35.25Improved
use entity-class-aware contrastive loss L2 (ablation)RelDN · OpenImages miniwmAP_rel: 34.63 -> 35.03Improved
use predicate-class-aware contrastive loss L3 (ablation)RelDN · OpenImages miniwmAP_rel: 34.63 -> 35.19Improved
use all three losses together instead of L1 and L2 only (ablation)RelDN · OpenImages miniAP_rel plays: 38.58 -> 41.04Improved
set every contrastive margin threshold to 1.0 instead of 0.2 (ablation)RelDN · OpenImages miniwmAP_rel: 35.54 -> 34.17Hurt
add visual subject-predicate-object features to the semantic module (ablation)RelDN · OpenImages miniscore_wtd: 39.32 -> 44.06Improved
add separate subject and object skip connections to the visual module (ablation)RelDN · OpenImages miniscore_wtd: 44.06 -> 44.44Improved
add the spatial module to semantic and visual modules (ablation)RelDN · OpenImages miniscore_wtd: 44.44 -> 44.61Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy