Referring Expression Segmentation on RefCOCOg (val-u)
83.8cIoUX-SAM + Phi3 3.8B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| X-SAM + Phi3 3.8BZero-Shot Transfer=false, Backbone/MLLM=Phi3 3.8B2026.04 | 83.8 | — | — | — | |
| HyperSegType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 79.4 | — | — | — | |
| HyperSeg + Phi2 2.7BZero-Shot Transfer=false, Backbone/MLLM=Phi2 2.7B2026.04 | 79.4 | — | — | — | |
| POPENFinetuning=true2025.04 | 75.4 | — | — | — | |
| SegAgent-Qwen+SClickBackbone=Qwen, Interactive Model=SimpleClick2025.03 | 75.11 | — | — | — | |
| SegAgent-LLaVA+SAMBackbone=LLaVA, Interactive Model=SAM2025.03 | 74.8 | — | — | — | |
| UGround-7BParameters=7B2025.10 | 74.7 | — | — | — | |
| SAM4MLLM-7BType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 74.5 | — | — | — | |
| SAM4MLLM(Qwen)Venue=ECCV'24, Backbone=Qwen, Interactive Model=SAM2025.03 | 74.5 | — | — | — | |
| SegAgent-Qwen+SAMBackbone=Qwen, Interactive Model=SAM2025.03 | 74.49 | — | — | — | |
| GLaMMType=MLLM-based Segmentation Network, SAM-based=true2024.11 | 74.2 | — | — | — | |
| GLaMM-7BParameters=7B2025.10 | 74.2 | — | — | — | |
| GroundHog-7BType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 74.1 | — | — | — | |
| PSALMType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 73.8 | — | — | — | |
| POPEN2025.04 | 73.8 | — | — | — | |
| Vision BananaZero-Shot Transfer=true2026.04 | 73.8 | — | — | — | |
| SAM 3 + Gemini 2.5 ProZero-Shot Transfer=true, Backbone/MLLM=Gemini 2.5 Pro2026.04 | 73.4 | — | — | — | |
| OMG-LLaVAType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 72.9 | — | — | — | |
| OMG-LLaVA2025.10 | 72.9 | — | — | — | |
| GSVA(SAM)Venue=CVPR'24, Interactive Model=SAM2025.03 | 72.7 | — | — | — | |
| GSVAFinetuning=true2025.04 | 72.7 | — | — | — | |
| GSVA-7BParameters=7B2025.10 | 72.7 | — | — | — | |
| SegLLM-7BParameters=7B2025.10 | 72.6 | — | — | — | |
| SETOKIM2024.06 | 71.3 | — | — | — | |
| PolyFormer-LVenue=CVPR'232025.03 | 71.15 | — | — | — | |
| GSVA-7BType=MLLM-based Segmentation Network, SAM-based=true2024.11 | 71.1 | — | — | — | |
| GSVA2025.04 | 71.1 | — | — | — | |
| LaSagnA-7BType=MLLM-based Segmentation Network, SAM-based=true2024.11 | 70.6 | — | — | — | |
| SegAgent-LLaVA+SClickBackbone=LLaVA, Interactive Model=SimpleClick2025.03 | 70.45 | — | — | — | |
| PerceptionGPTVenue=CVPR'242025.03 | 70.3 | — | — | — | |
| F-LMM-7BType=MLLM-based Segmentation Network, SAM-based=true2024.11 | 70.1 | — | — | — | |
| READ-7BParameters=7B2025.10 | 70.1 | — | — | — | |
| PixelLM-7BType=MLLM-based Segmentation Network, SAM-based=false2024.11 | 69.3 | — | — | — | |
| PixelLM2024.06 | 69.3 | — | — | — | |
| PixelLMVenue=CVPR'242025.03 | 69.3 | — | — | — | |
| PixelLM2025.04 | 69.3 | — | — | — | |
| PixelLM-7BParameters=7B2025.10 | 69.3 | — | — | — | |
| CoRes2025.10 | 69 | — | — | — | |
| LISA-7BFine-tuned on ReferSeg=true2023.08 | 67.9 | — | — | — | |
| LISABackbone=LLaVA-7B2024.07 | 67.9 | — | — | — | |
| LISA-7BType=MLLM-based Segmentation Network, SAM-based=true2024.11 | 67.9 | — | — | — | |
| LISA2024.06 | 67.9 | — | — | — | |
| LISA(SAM)Venue=CVPR'24, Interactive Model=SAM2025.03 | 67.9 | — | — | — | |
| LISAFinetuning=true2025.04 | 67.9 | — | — | — | |
| LISA-7BParameters=7B2025.10 | 67.9 | — | — | — | |
| PolyFormer-BType=Segmentation Specialist, SAM-based=false2024.11 | 67.8 | — | — | — | |
| Kang et al. + LLaVA-1.5 13BZero-Shot Transfer=true, Backbone/MLLM=LLaVA-1.5 13B2026.04 | 67.7 | — | — | — | |
| NEXT-Chat2024.06 | 67 | — | — | — | |
| Qwen box + SAMInteractive Model=SAM2025.03 | 66.93 | — | — | — | |
| LISA-7B2023.08 | 66.4 | — | — | — | |
| LISA2025.04 | 66.4 | — | — | — | |
| SESAME2025.10 | 66.1 | — | — | — | |
| SEEM2023.08 | 65.7 | — | — | — | |
| SEEMBackbone=DaViT-L2024.07 | 65.7 | — | — | — | |
| SEEM2024.06 | 65.7 | — | — | — | |
| SEEMVenue=NeurIPS'232025.03 | 65.7 | — | — | — | |
| SEEM2025.04 | 65.7 | — | — | — | |
| SEEM2025.10 | 65.7 | — | — | — | |
| VISABackbone=Chat-UniVi-7B2024.07 | 65.5 | — | — | — | |
| VISA-7BParameters=7B2025.10 | 65.5 | — | — | — | |
| ReLA2023.08 | 65 | — | — | — | |
| ReLABackbone=Swin-B2024.07 | 65 | — | — | — | |
| ReLA2024.06 | 65 | — | — | — | |
| ReLA2025.04 | 65 | — | — | — | |
| ReLA2025.10 | 65 | — | — | — | |
| X-Decoder2023.08 | 64.6 | — | — | — | |
| X-DecoderBackbone=DaViT-L2024.07 | 64.6 | — | — | — | |
| X-DecoderVenue=CVPR'232025.03 | 64.6 | — | — | — | |
| X-Decoder2025.04 | 64.6 | — | — | — | |
| X-Decoder2025.10 | 64.6 | — | — | — | |
| LISABackbone=LLaVA-7B, Note=trained via LISA's official GitHub repository2024.07 | 63.8 | — | — | — | |
| LAVT2023.08 | 61.2 | — | — | — | |
| LAVTBackbone=Swin-B2024.07 | 61.2 | — | — | — | |
| LAVTType=Segmentation Specialist, SAM-based=false2024.11 | 61.2 | — | — | — | |
| LAVTVenue=CVPR'222025.03 | 61.2 | — | — | — | |
| LAVT2025.04 | 61.2 | — | — | — | |
| LAVT2025.10 | 61.2 | — | — | — | |
| CRIS2023.08 | 59.9 | — | — | — | |
| CRISBackbone=ResNet1012024.07 | 59.9 | — | — | — | |
| CRISType=Segmentation Specialist, SAM-based=false2024.11 | 59.9 | — | — | — | |
| CRISVenue=CVPR'222025.03 | 59.9 | — | — | — | |
| CRIS2025.04 | 59.9 | — | — | — | |
| CRIS2025.10 | 59.9 | — | — | — | |
| Segment Anyword2025.10 | 58.4 | — | — | — | |
| VLT2023.08 | 55 | — | — | — | |
| VLTType=Segmentation Specialist, SAM-based=false2024.11 | 55 | — | — | — | |
| VLT2025.04 | 55 | — | — | — | |
| VLT2025.10 | 55 | — | — | — | |
| VLTBackbone=Darknet532024.07 | 53 | — | — | — | |
| HybridGLZero-Shot Transfer=true2026.04 | 51.3 | — | — | — | |
| MCN2023.08 | 49.2 | — | — | — | |
| MCNBackbone=Darknet532024.07 | 49.2 | — | — | — | |
| MCN2025.04 | 49.2 | — | — | — | |
| MCN2025.10 | 49.2 | — | — | — | |
| MAttNetVenue=CVPR'182025.03 | 47.64 | — | — | — | |
| B2GfixBackbone=BLIP2025.09 | — | 30.8 | — | — | |
| CaRmode=zero-shot2026.05 | — | — | 36.7 | — | |
| CoHDSize=-2026.01 | — | — | 70.83 | — | |
| CoHDSize=-2026.01 | — | — | 70.83 | — | |
| CRIS2023.11 | — | — | 59.9 | — |