Open vocabulary image classification on Wiki-H
934Used ObjectsNOVIC
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| NOVICSeen Objects=42919, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT0, Evaluation=Human annotations2024.07 | 934 | 28.7 | 0.771 | 0.7918 | 0.7821 | |
| NOVICSeen Objects=11897, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT2, Evaluation=Human annotations2024.07 | 893 | 36.3 | 0.7705 | 0.8013 | 0.7902 | |
| NOVICSeen Objects=42919, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT02024.07 | 852 | 37.5 | 0.687 | 0.7234 | 0.7131 | |
| NOVICSeen Objects=11897, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT22024.07 | 846 | 39.7 | 0.6865 | 0.7253 | 0.7188 | |
| NOVICSeen Objects=5899, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT62024.07 | 794 | 49.3 | 0.699 | 0.727 | 0.7107 | |
| NOVICSeen Objects=2919, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT92024.07 | 693 | 64 | 0.6855 | 0.7235 | 0.6908 | |
| RAMSeen Objects=42919, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT02024.07 | 399 | 81.7 | 0.617 | 0.7694 | 0.6025 | |
| Tag2TextSeen Objects=3429, Backbone=DFN-5B H/14-378 CLIP2024.07 | 397 | 89 | 0.6915 | 0.8135 | 0.6261 | |
| RAMSeen Objects=11897, Backbone=DFN-5B H/14-378 CLIP, Frequency Threshold=FT22024.07 | 391 | 83.4 | 0.623 | 0.7762 | 0.6058 | |
| RAMSeen Objects=4585, Backbone=DFN-5B H/14-378 CLIP2024.07 | 348 | 100 | 0.6505 | 0.8073 | 0.5998 |