Text-to-Image Retrieval on RSITMD (test)
25.04R@1GeoRSCLIP-FT
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GeoRSCLIP-FTParadigm=GVLM + FT, Tuned on=RS5M + RET-22023.06 | 25.04 | 57.88 | 74.38 | 51.81 | — | |
| GeoRSCLIP-FTParadigm=GVLM + FT, Tuned on=RS5M + RSITMD2023.06 | 23.54 | 57.52 | 74.6 | 50.1 | — | |
| ELCBackbone=ViT-B/322026.07 | 23.32 | 56.06 | 72.43 | — | 294.07 | |
| ELC0Backbone=ViT-B/322026.07 | 22.65 | 55.97 | 72.52 | — | 291.84 | |
| GLGSABackbone=ViT-B/322026.07 | 22.52 | 52.39 | 69.46 | — | 279.76 | |
| RemoteCLIPParadigm=GVLM + FT, Tuned on=RET-3 + DET-10 + SEG-42023.06 | 22.17 | 56.46 | 73.41 | 49.38 | — | |
| RemoteCLIPBackbone=ViT-B/322026.07 | 22.17 | 56.46 | 73.41 | — | 296.29 | |
| FGVLABackbone=ViT-B/322026.07 | 21.77 | 54.12 | 72.92 | — | 291.51 | |
| Full fine-tuningReference=-, Backbone (image/text)=CLIP(ViT-B-32), Params=151M2023.08 | 20.4 | 50.53 | 68.54 | 45.33 | — | |
| OursReference=-, Backbone (image/text)=CLIP(ViT-B-32), Params=0.16M2023.08 | 20.1 | 50.63 | 67.97 | 44.47 | — | |
| PE-RSITRParadigm=GVLM + FT, Tuned on=RSITMD2023.06 | 20.1 | 50.63 | 67.97 | 44.47 | — | |
| PE-RSITRBackbone=ViT-B/322026.07 | 20.1 | 50.63 | 67.97 | — | 266.8 | |
| SIRSBackbone=ViT-B/322026.07 | 19.44 | 43.18 | 58.4 | — | 241.26 | |
| AIRBackbone=ViT-B/322026.07 | 19.2 | 50.22 | 68.89 | — | 258.67 | |
| CUPBackbone=ViT-B/322026.07 | 17.7 | 48.4 | 64.82 | — | 252.94 | |
| Single LanguageReference=JSTARS’22, Backbone (image/text)=CLIP(ViT-B-32), Params=151M2023.08 | 17.61 | 49.73 | 66.59 | 41.38 | — | |
| MultilanguageParadigm=Supervised, Tuned on=RSITMD2023.06 | 17.61 | 49.73 | 66.59 | 41.38 | — | |
| UniAdapterReference=arXiv’23, Backbone (image/text)=CLIP(ViT-B-32), Params=0.55M2023.08 | 17.54 | 44.89 | 65.46 | 39.23 | — | |
| LRSCLIPBackbone=ViT-L/142026.03 | 17.39 | 41.59 | 56.77 | — | — | |
| GeoAlignCLIPBackbone=ViT-L/142026.03 | 17.39 | 42.39 | 58.05 | — | — | |
| InfoNCEBackbone=ViT-B/322026.07 | 17.26 | 47.88 | 67.61 | — | 245.14 | |
| GeoRSCLIPBackbone=ViT-L/142026.03 | 17.21 | 41.37 | 56.33 | — | — | |
| VSE∞Backbone=ViT-B/322026.07 | 16.73 | 47.21 | 61.37 | — | 250.09 | |
| MTGFEParadigm=Supervised, Tuned on=RSITMD2023.06 | 16.59 | 48.5 | 67.43 | 40.78 | — | |
| PAUBackbone=ViT-B/322026.07 | 16.37 | 49.2 | 66.46 | — | 250.39 | |
| Cross-Modal AdapterReference=arXiv’22, Backbone (image/text)=CLIP(ViT-B-32), Params=0.16M2023.08 | 16.31 | 44.33 | 64.75 | 38.06 | — | |
| GeoRSCLIPBackbone=ViT-L/14-3362026.03 | 16.28 | 42.08 | 57.39 | — | — | |
| VPTReference=ECCV’22, Backbone (image/text)=CLIP(ViT-B-32), Params=0.46M2023.08 | 15.97 | 41.35 | 60.35 | 34.14 | — | |
| FGCLIPBackbone=ViT-L/14-3362026.03 | 15.84 | 42.3 | 58.1 | — | — | |
| GeoAlignCLIPBackbone=ViT-B/162026.03 | 15 | 40.49 | 55.49 | — | — | |
| LRSCLIPBackbone=ViT-B/162026.03 | 14.65 | 37.21 | 51.99 | — | — | |
| FGCLIPBackbone=ViT-B/162026.03 | 14.6 | 38.32 | 52.92 | — | — | |
| KAMCL2026.07 | 14.6 | 41.86 | 59.86 | — | 215.21 | |
| VSE++Backbone=ViT-B/322026.07 | 14.47 | 41.86 | 56.81 | — | 223.54 | |
| AdaptFormerReference=NIPS’22, Backbone (image/text)=CLIP(ViT-B-32), Params=0.17M2023.08 | 14.27 | 41.53 | 61.46 | 34.81 | — | |
| GeoRSCLIPParadigm=GVLM + FT, Tuned on=RS5M2023.06 | 14.16 | 42.39 | 57.52 | 35.68 | — | |
| PolyBackbone=ViT-B/322026.07 | 13.98 | 41.9 | 60.09 | — | 220.62 | |
| Linear probeReference=ICML’21, Backbone (image/text)=CLIP(ViT-B-32), Params=1.05M2023.08 | 13.33 | 41.8 | 63.89 | 36.25 | — | |
| CLIP-AdapterReference=arXiv’21, Backbone (image/text)=CLIP(ViT-B-32), Params=0.52M2023.08 | 13.3 | 40.2 | 60.06 | 32.38 | — | |
| SMLGN2026.07 | 13.19 | 43.94 | 60.4 | — | 225.72 | |
| FAAMIParadigm=Supervised, Tuned on=RSITMD2023.06 | 12.96 | 42.39 | 59.96 | 35.99 | — | |
| GeoRSCLIPParadigm=GVLM + LoRA, Tuned on=RS5M2023.06 | 12.92 | 34.87 | 50.88 | 31.78 | — | |
| AdapterReference=ICML’19, Backbone (image/text)=CLIP(ViT-B-32), Params=0.17M2023.08 | 12.89 | 40.09 | 59.91 | 32.37 | — | |
| LongCLIPBackbone=ViT-L/142026.03 | 12.74 | 36.9 | 53.14 | — | — | |
| VL-PromptReference=-, Backbone (image/text)=CLIP(ViT-B-32), Params=0.47M2023.08 | 12.61 | 36.2 | 58.84 | 32.4 | — | |
| GeoRSCLIPBackbone=ViT-B/322026.03 | 12.35 | 38.58 | 55.31 | — | — | |
| GeoRSCLIPParadigm=GVLM + Pfeiffer, Tuned on=RS5M2023.06 | 12.26 | 38.1 | 53.94 | 33.42 | — | |
| DOVEParms=38 M2023.10 | 12.2 | 44.13 | 66.5 | — | — | |
| PIRParadigm=Supervised, Tuned on=RSITMD2023.06 | 12.17 | 41.68 | 63.41 | 38.24 | — | |
| DOVE-SParms=25 M2023.10 | 12.09 | 42.39 | 59.16 | — | — | |
| ConvpassReference=arXiv’22, Backbone (image/text)=CLIP(ViT-B-32), Params=0.14M2023.08 | 12.05 | 38.66 | 58.11 | 32.55 | — | |
| GeoRSCLIPParadigm=GVLM + UniPELT, Tuned on=RS5M2023.06 | 11.68 | 36.15 | 52.7 | 31.76 | — | |
| AMFMN-simReference=TGRS’22, Backbone (image/text)=ResNet18, biGRU, Params=35.94M2023.08 | 11.51 | 34.69 | 54.87 | 29.72 | — | |
| AMFMNParms=36 M2023.10 | 11.51 | 34.69 | 54.87 | — | — | |
| AMFMN2026.07 | 11.51 | 34.69 | 54.87 | — | 178.29 | |
| HVSAParadigm=Supervised, Tuned on=RSITMD2023.06 | 11.43 | 39.2 | 57.45 | 33.15 | — | |
| HVSA2023.10 | 11.43 | 39.2 | 57.45 | — | — | |
| GeoRSCLIPParadigm=GVLM + Prefixtuning, Tuned on=RS5M2023.06 | 11.42 | 34.2 | 50.88 | 31.02 | — | |
| SWANParadigm=Supervised, Tuned on=RSITMD2023.06 | 11.24 | 40.4 | 60.6 | 34.11 | — | |
| SWANParms=40 M2023.10 | 11.24 | 40.4 | 60.6 | — | — | |
| SkyCLIPBackbone=ViT-L/14-pct302026.03 | 11.19 | 34.87 | 50.4 | — | — | |
| GaLR with MRReference=TGRS’22, Backbone (image/text)=ResNet18, biGRU, Params=46.89M2023.08 | 11.15 | 36.68 | 51.68 | 31.41 | — | |
| GaLRParadigm=Supervised, Tuned on=RSITMD2023.06 | 11.15 | 36.68 | 51.68 | 31.41 | — | |
| GaLRParms=46 M2023.10 | 11.15 | 36.68 | 51.68 | — | — | |
| GaLR2026.07 | 11.15 | 36.68 | 51.68 | — | 188.45 | |
| SkyCLIPBackbone=ViT-L/14-pct502026.03 | 10.75 | 34.69 | 50.58 | — | — | |
| GaLR w/o MRReference=TGRS’22, Backbone (image/text)=ResNet18, biGRU, Params=46.89M2023.08 | 10.47 | 36.34 | 53.35 | 31 | — | |
| SkyCLIPBackbone=ViT-B/322026.03 | 10.31 | 33.63 | 49.6 | — | — | |
| SCAN t-iReference=ECCV’18, Backbone (image/text)=Faster R-CNN, biGRU, Params=13.68M2023.08 | 10.1 | 28.98 | 43.53 | 26.64 | — | |
| CLIPBackbone=ViT-L/142026.03 | 10.04 | 31.33 | 46.15 | — | — | |
| MTFNReference=ACM MM’19, Backbone (image/text)=Faster R-CNN, biGRU, Params=77.90M2023.08 | 9.96 | 31.37 | 45.84 | 26.92 | — | |
| AMFMN-fusionReference=TGRS’22, Backbone (image/text)=ResNet18, biGRU, Params=35.94M2023.08 | 9.96 | 34.03 | 52.96 | 29.32 | — | |
| AFMFNParadigm=Supervised, Tuned on=RSITMD2023.06 | 9.96 | 34.03 | 52.96 | 29.32 | — | |
| SCAN i-tReference=ECCV’18, Backbone (image/text)=Faster R-CNN, biGRU, Params=13.68M2023.08 | 9.82 | 29.38 | 42.12 | 26.28 | — | |
| AMFMN-softReference=TGRS’22, Backbone (image/text)=ResNet18, biGRU, Params=35.94M2023.08 | 9.82 | 33.94 | 51.9 | 28.74 | — | |
| MCRNReference=JAG’22, Backbone (image/text)=ResNet18, biGRU, Params=52.35M2023.08 | 9.42 | 35.53 | 52.74 | 30.33 | — | |
| LW-MCR-uReference=TGRS’22, Backbone (image/text)=SqueezeNet, -, Params=1.65M2023.08 | 9.25 | 34.07 | 54.03 | 28.58 | — | |
| CoOpReference=IJCV’22, Backbone (image/text)=CLIP(ViT-B-32), Params=0.10M2023.08 | 9.16 | 33.85 | 54.35 | 30.51 | — | |
| CLIP-BaselineParadigm=GVLM, Tuned on=None2023.06 | 8.81 | 27.88 | 43.19 | 24.19 | — | |
| CLIPBackbone=ViT-B/322026.07 | 8.81 | 28.01 | 43.1 | — | 145.4 | |
| LongCLIPBackbone=ViT-B/162026.03 | 8.72 | 33.41 | 49.07 | — | — | |
| CLIPBackbone=ViT-B/162026.03 | 8.36 | 26.11 | 40.84 | — | — | |
| CAMP-tripletReference=ICCV’19, Backbone (image/text)=Faster R-CNN, biGRU, Params=27.03M2023.08 | 8.27 | 27.79 | 44.34 | 26.2 | — | |
| VSE++Reference=BMVC’18, Backbone (image/text)=ResNet18, GRU, Params=15.78M2023.08 | 7.79 | 24.87 | 38.67 | 24.83 | — | |
| LW-MCR-dReference=TGRS’22, Backbone (image/text)=SqueezeNet, -, Params=1.65M2023.08 | 7.79 | 30.18 | 49.78 | 27.79 | — | |
| Zero-shot CLIPReference=ICML’21, Backbone (image/text)=CLIP(ViT-B-32), Params=0.00M2023.08 | 7.79 | 23.67 | 38.89 | 23.89 | — | |
| LW-MCRParadigm=Supervised, Tuned on=RSITMD2023.06 | 7.79 | 30.18 | 49.78 | 27.79 | — | |
| VSE++Paradigm=Supervised, Tuned on=RSITMD2023.06 | 7.79 | 24.87 | 38.67 | 24.83 | — | |
| LW-MCR2023.10 | 7.79 | 30.18 | 49.78 | — | — | |
| VSE++Parms=29 M2023.10 | 7.73 | 27.8 | 41 | — | — | |
| CAMERAParms=64 M2023.10 | 7.52 | 26.19 | 40.72 | — | — | |
| SCAN i2tParms=60 M2023.10 | 7.43 | 25.71 | 39.03 | — | — | |
| SCAN t2iParms=60 M2023.10 | 7.06 | 26.49 | 42.21 | — | — | |
| CAMPParms=63 M2023.10 | 6.24 | 26.37 | 42.37 | — | — | |
| LW-MCR-bReference=TGRS’22, Backbone (image/text)=SqueezeNet, -, Params=1.65M2023.08 | 6.11 | 27.74 | 49.56 | 25.55 | — | |
| CAMP-bceReference=ICCV’19, Backbone (image/text)=Faster R-CNN, biGRU, Params=27.03M2023.08 | 5.22 | 23.32 | 38.36 | 22.03 | — |