Video Object Segmentation on SA-V (val)
83.5J&F ScoreSAM3
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SAM3Mode=One-by-one2026.01 | 83.5 | — | — | — | |
| SAM3*Mode=One-by-one2026.01 | 83.3 | 79.4 | 87.2 | — | |
| SAM3-DMSMode=Simultaneous2026.01 | 83.3 | 79.4 | 87.2 | — | |
| SeC2025.07 | 82.7 | — | — | — | |
| SENTRY-HiM-LEvaluation Protocol=first-frame-mask2026.06 | 82.4 | — | — | — | |
| SENTRY-SA-LEvaluation Protocol=first-frame-mask2026.06 | 82.1 | — | — | — | |
| SENTRY-D4S-LEvaluation Protocol=first-frame-mask2026.06 | 81.8 | — | — | — | |
| SAM3Mode=Simultaneous2026.01 | 81.2 | 77.4 | 84.9 | — | |
| SAM2.1LongModel Scale=L2024.10 | 81.1 | — | — | — | |
| SAM2.1Long2024.10 | 81.1 | 77.5 | 84.7 | — | |
| SAM2LongSize=Large, Base Version=2.12024.10 | 81.1 | 77.5 | 84.7 | — | |
| SAM2.1LongMode=Long-term2025.07 | 81.1 | — | — | — | |
| SAM2Long-LEvaluation Protocol=first-frame-mask2026.06 | 81.1 | — | — | — | |
| HiM2SAM-LEvaluation Protocol=first-frame-mask2026.06 | 81 | — | — | — | |
| SAM2Long2024.10 | 80.8 | 74.7 | 84.7 | — | |
| SAM2LongSize=Large, Base Version=2.02024.10 | 80.8 | 77.1 | 84.5 | — | |
| SAMITE-LEvaluation Protocol=first-frame-mask2026.06 | 80.8 | — | — | — | |
| SENTRY-SR-LEvaluation Protocol=first-frame-mask2026.06 | 80.6 | — | — | — | |
| SAM2LongSize=Base+, Base Version=2.12024.10 | 80.5 | 76.8 | 84.2 | — | |
| DAM4SAM-LEvaluation Protocol=first-frame-mask2026.06 | 80.5 | — | — | — | |
| SAMURAI2025.07 | 79.8 | — | — | — | |
| OriginalMemory interval (delta_t)=52026.02 | 79.7 | — | — | — | |
| SAM2LongSize=Small, Base Version=2.12024.10 | 79.6 | 75.9 | 83.3 | — | |
| SAMURAI-LEvaluation Protocol=first-frame-mask2026.06 | 79.5 | — | — | — | |
| SENTRY-S2-LEvaluation Protocol=first-frame-mask2026.06 | 79.4 | — | — | — | |
| SAM2LongSize=Tiny, Base Version=2.12024.10 | 78.9 | 75.2 | 82.7 | — | |
| SMRIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.78x, Memory interval (delta_t)=52026.02 | 78.8 | — | — | — | |
| SAM 2.1Model Scale=L2024.10 | 78.6 | — | — | — | |
| SAM 2.1+2024.10 | 78.6 | 75.1 | 82 | — | |
| SAM 2.1Size=Large, Base Version=2.12024.10 | 78.6 | 75.1 | 82 | — | |
| SAM 2.12025.07 | 78.6 | — | — | — | |
| SAM2-LEvaluation Protocol=first-frame-mask2026.06 | 78.6 | — | — | — | |
| SAM2LongSize=Base+, Base Version=2.02024.10 | 78.4 | 74.7 | 82.1 | — | |
| OriginalMemory interval (delta_t)=12026.02 | 78.3 | — | — | — | |
| SMR-randomIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.75x, Memory interval (delta_t)=52026.02 | 78.2 | — | — | — | |
| SMR-uniformIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.76x, Memory interval (delta_t)=52026.02 | 78.2 | — | — | — | |
| SAM 2.1Size=Base+, Base Version=2.12024.10 | 78 | 74.6 | 81.5 | — | |
| SAM 2 (Hiera-L)Backbone=Hiera-L, Prompting protocol=first-frame ground-truth mask2024.08 | 77.9 | — | — | — | |
| SAM2LongSize=Small, Base Version=2.02024.10 | 77.7 | 73.9 | 81.5 | — | |
| SWRIE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=false, Speedup=1.83x, Memory interval (delta_t)=52026.02 | 77.7 | — | — | — | |
| SMR-uniformIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.76x, Memory interval (delta_t)=12026.02 | 77.7 | — | — | — | |
| SMRIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.78x, Memory interval (delta_t)=12026.02 | 77.7 | — | — | — | |
| Efficient-SAM2IE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=true, Speedup=1.80x, Memory interval (delta_t)=52026.02 | 77.7 | — | — | — | |
| SAM2LongSize=Tiny, Base Version=2.02024.10 | 77 | 73.2 | 80.7 | — | |
| SAM 2.1Size=Small, Base Version=2.12024.10 | 76.9 | 73.5 | 80.3 | — | |
| SAM 2 (Hiera-B+)Backbone=Hiera-B+, Prompting protocol=first-frame ground-truth mask2024.08 | 76.8 | — | — | — | |
| SAM 2.1-B+2025.01 | 76.8 | — | — | — | |
| G2TAM2026.07 | 76.8 | — | — | — | |
| SMR-randomIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=1.75x, Memory interval (delta_t)=12026.02 | 76.6 | — | — | — | |
| SWRIE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=false, Speedup=1.83x, Memory interval (delta_t)=12026.02 | 76.4 | — | — | — | |
| SAM2-Base+ (Reference)Model=SAM2-Base+, Precision=FP2025.03 | 76.35 | — | — | — | |
| SAM 2Size=Large, Base Version=2.02024.10 | 76.3 | 73 | 79.5 | — | |
| SAM 22024.10 | 76.1 | 72.9 | 79.2 | — | |
| SAM2-Small (Reference)Model=SAM2-Small, Precision=FP2025.03 | 76.1 | — | — | — | |
| SAM 22026.07 | 75.8 | — | — | — | |
| Efficient-SAM2IE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=true, Speedup=1.80x, Memory interval (delta_t)=12026.02 | 75.5 | — | — | — | |
| SAM 2Size=Base+, Base Version=2.02024.10 | 75.4 | 71.9 | 78.8 | — | |
| SAM 2.1Size=Tiny, Base Version=2.12024.10 | 75.1 | 71.6 | 78.6 | — | |
| MemPoolIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=2.04x, Memory interval (delta_t)=52026.02 | 74.9 | — | — | — | |
| SAM2-Tiny (Reference)Model=SAM2-Tiny, Precision=FP2025.03 | 74.48 | — | — | — | |
| SAM 2-B+2025.01 | 73.6 | — | — | — | |
| SAM 2Size=Tiny, Base Version=2.02024.10 | 73.5 | 70.1 | 76.9 | — | |
| MemPoolIE (Image Encoder Optimization)=false, MA (Memory Attention Optimization)=true, Speedup=2.04x, Memory interval (delta_t)=12026.02 | 73.2 | — | — | — | |
| SAM 2Size=Small, Base Version=2.02024.10 | 73 | 69.7 | 76.3 | — | |
| ToMeIE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=false, Speedup=1.43x, Memory interval (delta_t)=12026.02 | 72.4 | — | — | — | |
| EdgeTAM2025.01 | 72.3 | — | — | — | |
| AHCQ-SAMModel=SAM2-Tiny, Precision=W6A62025.03 | 72.12 | — | — | — | |
| ToMe4DMIE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=false, Speedup=1.14x, Memory interval (delta_t)=52026.02 | 71.6 | — | — | — | |
| AHCQ-SAMModel=SAM2-Small, Precision=W6A62025.03 | 70.85 | — | — | — | |
| PTQ4SAMModel=SAM2-Tiny, Precision=W6A62025.03 | 70.65 | — | — | — | |
| PTQ4SAMModel=SAM2-Small, Precision=W6A62025.03 | 70.45 | — | — | — | |
| ToMe4DMIE (Image Encoder Optimization)=true, MA (Memory Attention Optimization)=false, Speedup=1.14x, Memory interval (delta_t)=12026.02 | 69.6 | — | — | — | |
| AHCQ-SAMModel=SAM2-Base+, Precision=W6A62025.03 | 64.69 | — | — | — | |
| PTQ4SAMModel=SAM2-Base+, Precision=W6A62025.03 | 64.03 | — | — | — | |
| QDropModel=SAM2-Small, Precision=W6A62025.03 | 63.82 | — | — | — | |
| QDropModel=SAM2-Tiny, Precision=W6A62025.03 | 62.34 | — | — | — | |
| SwinB-DeAOT2024.10 | 61.4 | 56.6 | 66.2 | — | |
| SwinB-DeAOTBackbone=SwinB, Prompting protocol=first-frame ground-truth mask2024.08 | 61.4 | — | — | — | |
| SwinB-DeAOT2025.01 | 61.4 | — | — | — | |
| SwinB-DeAOTBackbone=Swin-B2025.07 | 61.4 | — | — | — | |
| SwinB-DeAOTBackbone=SwinB2026.07 | 61.4 | — | — | — | |
| Cutie-base+2024.10 | 61.3 | 58.3 | 64.4 | — | |
| Cutie-base+Backbone=base+, Prompting protocol=first-frame ground-truth mask2024.08 | 61.3 | — | — | — | |
| Cutie-base+2025.01 | 61.3 | — | — | — | |
| Cutie-base+Config=base+2025.07 | 61.3 | — | — | — | |
| Cutie-base+Variant=base+2026.07 | 61.3 | — | — | — | |
| STCN2024.10 | 61 | 57.4 | 64.5 | — | |
| STCNPrompting protocol=first-frame ground-truth mask2024.08 | 61 | — | — | — | |
| STCN2025.01 | 61 | — | — | — | |
| STCN2025.07 | 61 | — | — | — | |
| STCN2026.07 | 61 | — | — | — | |
| Cutie-baseBackbone=base, Prompting protocol=first-frame ground-truth mask2024.08 | 60.7 | — | — | — | |
| Cutie-base2025.01 | 60.7 | — | — | — | |
| Cutie-baseConfig=base2025.07 | 60.7 | — | — | — | |
| Cutie-baseVariant=base2026.07 | 60.7 | — | — | — | |
| XMem2024.10 | 60.1 | 56.3 | 63.9 | — | |
| XMemPrompting protocol=first-frame ground-truth mask2024.08 | 60.1 | — | — | — | |
| XMem2025.01 | 60.1 | — | — | — | |
| XMem2025.07 | 60.1 | — | — | — | |
| XMem2026.07 | 60.1 | — | — | — |