Referring Expression Segmentation on refCOCO UMD (val)
85.1cIoUX-SAM
Evaluation Results
| Method | Links | |
|---|---|---|
| X-SAMVersion=Phi3 3.8B, Training RES=true2025.12 | 85.1 | |
| HyperSegVersion=Phi2 2.7B, Training RES=true2025.12 | 84.8 | |
| LENSVersion=Qwen2.5-VL 3B, Training RES=true2025.12 | 84.2 | |
| PSALMVersion=Phi1.5 1.3B, Training RES=true2025.12 | 83.6 | |
| MLCD-segVersion=Qwen2.5-7B, Training RES=true2025.12 | 83.6 | |
| RICEVersion=Qwen2.5-7B, Training RES=true2025.12 | 83.5 | |
| EVF-SAMVersion=Extra Data, Training RES=true2025.12 | 82.4 | |
| EVF-SAMVersion=RC, Training RES=true2025.12 | 82.1 | |
| UniLSegVersion=UniLSeg-100, Training RES=true2025.12 | 81.7 | |
| DETRISVersion=DETRIS-L, Training RES=true2025.12 | 81 | |
| UniLSegVersion=UniLSeg-20, Training RES=true2025.12 | 80.5 | |
| GLEEVersion=Pro, Training RES=true2025.12 | 80 | |
| SAM4MLLMVersion=LLaVA1.6 8B, Training RES=true2025.12 | 79.8 | |
| SegAgent-Qwen+SClickBackbone=Qwen, Interactive Model=SimpleClick2025.03 | 79.69 | |
| SAM4MLLMVersion=LLaVA1.6 7B, Training RES=true2025.12 | 79.6 | |
| GLaMMVersion=Vicuna 7B, Training RES=true2025.12 | 79.5 | |
| GLEEVersion=Plus, Training RES=true2025.12 | 79.5 | |
| SegAgent-LLaVA+SAMBackbone=LLaVA, Interactive Model=SAM2025.03 | 79.2 | |
| GSVAVersion=13B, Training RES=true2025.12 | 79.2 | |
| SegAgentVersion=Qwen-VL 7B, Training RES=true2025.12 | 79.2 | |
| READVersion=LLaVA1.5 7B, Training RES=true2025.12 | 78.1 | |
| SegAgent-Qwen+SAMBackbone=Qwen, Interactive Model=SAM2025.03 | 78.01 | |
| SegAgent-LLaVA+SClickBackbone=LLaVA, Interactive Model=SimpleClick2025.03 | 77.81 | |
| GSVA(SAM)Venue=CVPR'24, Interactive Model=SAM2025.03 | 77.2 | |
| SAM4MLLM(Qwen)Venue=ECCV'24, Backbone=Qwen, Interactive Model=SAM2025.03 | 77.1 | |
| PolyFormer-LVenue=CVPR'232025.03 | 76.94 | |
| PerceptionGPTVenue=CVPR'242025.03 | 75.1 | |
| LISA(SAM)Venue=CVPR'24, Interactive Model=SAM2025.03 | 74.9 | |
| LISAVersion=LLaVA 7B, Training RES=true2025.12 | 74.9 | |
| PixelLMVenue=CVPR'242025.03 | 73 | |
| LAVTVenue=CVPR'222025.03 | 72.7 | |
| Qwen box + SAMInteractive Model=SAM2025.03 | 71.79 | |
| CRISVenue=CVPR'222025.03 | 70.5 | |
| EVOL-SAM3Version=Qwen2.5-VL 7B, Training RES=false2025.12 | 68.7 | |
| EVOL-SAM3Version=Qwen2.5-VL 3B, Training RES=false2025.12 | 66.8 | |
| SAM 3 AgentVersion=Qwen2.5-VL 7B, Training RES=false2025.12 | 59.4 | |
| MAttNetVenue=CVPR'182025.03 | 56.51 | |
| VLM-VGVersion=ResNet-101, Training RES=false2025.12 | 49.9 | |
| LGD+DINOVersion=ViT-B/32, Training RES=false2025.12 | 49.5 | |
| HybridGLVersion=ViT-B/32, Training RES=false2025.12 | 49.5 | |
| VLM-VGVersion=ResNet-50, Training RES=false2025.12 | 47.7 | |
| Pseudo-RISVersion=ETRIS, Training RES=false2025.12 | 41.1 | |
| IteRPrimeEVersion=N/A, Training RES=false2025.12 | 40.2 | |
| TASVersion=ResNet-50, Training RES=false2025.12 | 39.9 | |
| TASVersion=ViT-B/32, Training RES=false2025.12 | 39.8 | |
| Pseudo-RISVersion=CRIS, Training RES=false2025.12 | 39.8 | |
| Ref-DiffVersion=VAE, Training RES=false2025.12 | 37.2 | |
| CaRVersion=ViT-B/16, Training RES=false2025.12 | 33.6 | |
| GL-CLIPVersion=ViT-B/32, Training RES=false2025.12 | 32.9 | |
| GL-CLIPVersion=ResNet-50, Training RES=false2025.12 | 32.7 |