List your company
ResearchFormerly WizWand

Instance Segmentation on COCO (APm)

45APmFull-Precision
33.76836.68439.642.516Apr 3, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Full-PrecisionMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=32/322025.0445
Full-PrecisionMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=32/322025.0443.7
DopQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=TanQ2025.0443.7
APHQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0443.7
ERQMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=log √22025.0443.6
I&S-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=SULQ2025.0443.6
Full-PrecisionMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=32/322025.0443.3
RepQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=log √22025.0443.1
APHQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0442.7
QDropMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0442.5
ERQMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=log √22025.0442.1
DopQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=TanQ2025.0442.1
I&S-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=SULQ2025.0442
Full-PrecisionMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=32/322025.0441.6
BaselineMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=Uniform2025.0441.6
QDropMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0441.3
RepQ-ViTMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=log √22025.0441.1
APHQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0441
ERQMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=log √22025.0440.7
DopQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=TanQ2025.0440.4
I&S-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=SULQ2025.0440.3
BaselineMeta-Architecture=Cascade Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=Uniform2025.0440.2
RepQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=log √22025.0440.2
QDropMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0439.2
BaselineMeta-Architecture=Mask R-CNN, Backbone=Swin-S, W/A=4/4, Optimization=false, Quantizer=Uniform2025.0438.6
APHQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0438.1
ERQMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=log √22025.0436.6
I&S-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=SULQ2025.0436.6
DopQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=TanQ2025.0436.5
RepQ-ViTMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=log √22025.0436
QDropMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=true, Quantizer=Uniform2025.0435.4
BaselineMeta-Architecture=Mask R-CNN, Backbone=Swin-T, W/A=4/4, Optimization=false, Quantizer=Uniform2025.0434.2