Phrase Grounding On Flickr30K
Metrics
Pointing Game Accuracy
Results
Performance results of various models on this benchmark
| Paper Title | ||
|---|---|---|
| GBS Ensemble + 12-in-1 | 85.9 | Detector-Free Weakly Supervised Grounding by Separation |
| GbS Ensemble MS-COCO | 75.6 | Detector-Free Weakly Supervised Grounding by Separation |
| COCO_ELMo_PNASNet | 69.19 | Multi-level Multimodal Common Semantic Space for Image-Phrase Grounding |
0 of 3 row(s) selected.