Remote Sensing Change Captioning on LEVIR-CC
67.93BLEU-4TISDNet
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| TISDNetParameter (M)=91.40, GPU=1080Ti, Batch Size=12026.07 | 67.93 | — | — | — | — | 11.61 | |
| LBTCapParameter (M)=39.85, GPU=1080Ti, Batch Size=12026.07 | 65.94 | — | — | — | — | 18.94 | |
| LBTCapParameter (M)=39.85, GPU=4090, Batch Size=12026.07 | 65.94 | — | — | — | — | 62.26 | |
| LBTCapParameter (M)=39.85, GPU=3090, Batch Size=12026.07 | 65.94 | — | — | — | — | 39.28 | |
| LBTCapParameter (M)=39.85, GPU=A100, Batch Size=12026.07 | 65.94 | — | — | — | — | 27.21 | |
| SACNetModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 65.57 | 40.3 | 75.68 | 138.34 | 79.97 | — | |
| SAGE-CCModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 65.5 | 39.92 | 74.77 | 137.5 | 79.42 | — | |
| CTMTNetModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 64.69 | 39.49 | 74.54 | 134.94 | 78.42 | — | |
| SENModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 64.09 | 39.59 | 71.5 | 125.02 | 75.05 | — | |
| PromptCCModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 63.54 | 38.82 | 73.72 | 136.44 | 78.13 | — | |
| Chg2CapModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 62.98 | 39.42 | 74.34 | 136.25 | 78.25 | — | |
| RSICCFormer-CModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 62.41 | 38.7 | 73.6 | 132.62 | 76.83 | — | |
| PSNetModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 62.11 | 38.8 | 73.6 | 132.62 | 76.78 | — | |
| Qwen3.5 4B (SFT) + SVEModel Architecture Type (Specialist vs Generalist)=Generalist VLMs, Supervised Fine-Tuning (SFT)=true2026.06 | 61.33 | 39.91 | 76.26 | 144.35 | 80.46 | — | |
| Diffusion-RSCCModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 60.9 | 37.8 | 71.5 | 125.6 | 73.95 | — | |
| Qwen3.5 4B (SFT)Model Architecture Type (Specialist vs Generalist)=Generalist VLMs, Supervised Fine-Tuning (SFT)=true2026.06 | 60.7 | 39.42 | 76.03 | 142.26 | 79.6 | — | |
| DUDAModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 57.79 | 37.15 | 71.04 | 124.32 | 72.58 | — | |
| MCCFormer-DModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 56.38 | 39.91 | 70.44 | 124.44 | 72.79 | — | |
| MCCFormer-SModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 56.36 | 39.6 | 69.46 | 120.39 | 71.45 | — | |
| Capt-Att-Dual-AttModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 53.56 | 37.16 | 69.19 | 124.42 | 71.08 | — | |
| Capt-RepModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 53.15 | 36.58 | 69.73 | 121.22 | 70.17 | — | |
| Capt-DiffModel Architecture Type (Specialist vs Generalist)=Specialist models & architectures2026.06 | 47.41 | 34.47 | 65.64 | 110.57 | 64.52 | — | |
| Change3DParameter (M)=5.05, GPU=3090, Batch Size=12026.07 | — | — | — | — | — | 9.47 | |
| Chg2CapParameter (M)=285.5, GPU=A100, Batch Size=12026.07 | — | — | — | — | — | 11.98 | |
| FST-NetParameter (M)=111.76, GPU=4090, Batch Size=12026.07 | — | — | — | — | — | 31.35 | |
| ICT-NetParameter (M)=96.40, GPU=1080Ti, Batch Size=12026.07 | — | — | — | — | — | 10.44 | |
| Prompt-CCParameter (M)=408.58, GPU=4090, Batch Size=12026.07 | — | — | — | — | — | 9.94 | |
| RSICCformerParameter (M)=81.51, GPU=1080Ti, Batch Size=12026.07 | — | — | — | — | — | 12.4 | |
| SENParameter (M)=39.90, GPU=3090, Batch Size=12026.07 | — | — | — | — | — | 23.7 | |
| TACCParameter (M)=94.14, GPU=4090, Batch Size=12026.07 | — | — | — | — | — | 0.94 |