List your company
ResearchFormerly WizWand

Referring Expression Segmentation on RefCOCOg UMD (val)

69.5mIoULISA
-1.68816.793535.27553.7565Jul 2, 2024Oct 25, 2024Feb 18, 2025Jun 14, 2025Oct 7, 2025Jan 31, 2026May 27, 2026
Updated 5mo ago

Evaluation results

MethodLinks
LISATraining Mask Dec.=true, Learning Protocol=Supervised Fine-Tuning2025.1269.5
GETok-SFTTraining Mask Dec.=false, Learning Protocol=Supervised Fine-Tuning2025.1269.4
PixelLMTraining Mask Dec.=true, Learning Protocol=Supervised Fine-Tuning2025.1269.3
Qwen2.5-VL-7BTraining Mask Dec.=false, Learning Protocol=Supervised Fine-Tuning2025.1268.1
GETok-SFT-gridTraining Mask Dec.=false, Learning Protocol=Supervised Fine-Tuning2025.1268
ReLATraining Mask Dec.=true, Learning Protocol=Supervised Fine-Tuning2025.1265
SAFARILabel-Rate Mask=100%, Label-Rate BBox=100%, Vis. Backbone=Swin-B, Supervision Type=Fully Supervised2024.0762.42
LAVTTraining Mask Dec.=true, Learning Protocol=Supervised Fine-Tuning2025.1261.2
Learning to LabelLabel Budget=10%2026.0560.63
CRISTraining Mask Dec.=true, Learning Protocol=Supervised Fine-Tuning2025.1259.9
Baseline (w/ MLLM)Label Budget=10%2026.0559.74
Learning to LabelLabel Budget=5%2026.0559.71
Baseline (w/ MLLM)Label Budget=5%2026.0557.74
SAFARILabel-Rate Mask=30%, Label-Rate BBox=30%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0755.72
SeqTRLabel-Rate Mask=100%, Label-Rate BBox=100%, Vis. Backbone=DN53, Supervision Type=Fully Supervised2024.0755.67
Partial-RESLabel-Rate Mask=30%, Label-Rate BBox=100%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0754.69
SAFARILabel-Rate Mask=20%, Label-Rate BBox=20%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0754.45
LTSLabel-Rate Mask=100%, Label-Rate BBox=100%, Vis. Backbone=DN53, Supervision Type=Fully Supervised2024.0754.4
Partial-RESLabel-Rate Mask=20%, Label-Rate BBox=100%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0753.94
Learning to LabelLabel Budget=1%2026.0553.83
VLTLabel-Rate Mask=100%, Label-Rate BBox=100%, Vis. Backbone=DN53, Supervision Type=Fully Supervised2024.0752.99
SAFARILabel-Rate Mask=10%, Label-Rate BBox=10%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0752.91
Partial-RESLabel-Rate Mask=10%, Label-Rate BBox=100%, Vis. Backbone=Swin-B, Supervision Type=Weakly Supervised2024.0752.73
Baseline (w/ MLLM)Label Budget=1%2026.0551.97
Learning to LabelLabel Budget=0.1%2026.0551.01
Baseline (w/ MLLM)Label Budget=0.1%2026.0550.13
Baseline (w/o MLLM)Label Budget=10%2026.0549.26
SemiRESLabel Budget=5%2026.0547.61
Baseline (w/o MLLM)Label Budget=5%2026.0546.45
FixmatchLabel Budget=10%2026.0545.5
RESMatchLabel Budget=10%2026.0545.24
CCVCLabel Budget=10%2026.0542.5
SupervisedLabel Budget=10%2026.0540.43
RESMatchLabel Budget=5%2026.0539.77
FixmatchLabel Budget=5%2026.0539.55
CCVCLabel Budget=5%2026.0537.76
AugSegLabel Budget=10%2026.0537.02
Baseline (w/o MLLM)Label Budget=1%2026.0534.96
SemiRESLabel Budget=1%2026.0534.76
SupervisedLabel Budget=5%2026.0534.58
AugSegLabel Budget=5%2026.0534.08
RESMatchLabel Budget=1%2026.0532.18
FixmatchLabel Budget=1%2026.0531.62
CCVCLabel Budget=1%2026.0529.35
SupervisedLabel Budget=1%2026.0526.23
AugSegLabel Budget=1%2026.0524.07
Baseline (w/o MLLM)Label Budget=0.1%2026.0521.25
RESMatchLabel Budget=0.1%2026.0520.97
FixmatchLabel Budget=0.1%2026.0519.96
SupervisedLabel Budget=0.1%2026.0516.97
CCVCLabel Budget=0.1%2026.0511.58
AugSegLabel Budget=0.1%2026.051.05