List your company
ResearchFormerly WizWand

Semantic Segmentation on ADE20K 150 semantic categories (val)

51.6mIoUSPANetV2-S36-hybrid
25.632.3539.145.85Aug 1, 2024Sep 10, 2024Oct 20, 2024Nov 30, 2024Jan 9, 2025Feb 18, 2025Mar 31, 2025
Updated 5mo ago

Evaluation results

MethodLinks
SPANetV2-S36-hybridFramework=UperNet, Params (M)=71, MACs (G)=1004, FPS=11.9, Resolution=512 x 20482025.0351.6
CAFormer-S36Framework=UperNet, Params (M)=67, MACs (G)=1197, FPS=20.6, Resolution=512 x 20482025.0350.8
ConvFormer-S36Framework=UperNet, Params (M)=67, MACs (G)=1003, FPS=23.0, Resolution=512 x 20482025.0350.7
Conv2Former-SFramework=UperNet, Params (M)=79, Resolution=512 x 20482025.0350.3
SPANetV2-S36-pureFramework=UperNet, Params (M)=75, MACs (G)=1012, FPS=10.8, Resolution=512 x 20482025.0349.8
ConvNeXt-BFramework=UperNet, #Param.=122M, FLOPs=1170G, Input size=512 x 20482024.0949.6
ConvNeXt-SFramework=UperNet, Params (M)=82, MACs (G)=1027, FPS=27.7, Resolution=512 x 20482025.0349.6
Swin-BFramework=UperNet, #Param.=121M, FLOPs=1188G, Input size=512 x 20482024.0949.5
Swin-SFramework=UperNet, Params (M)=81, MACs (G)=1038, FPS=28.7, Resolution=512 x 20482025.0349.5
SPANetV2-S18-hybridFramework=UperNet, Params (M)=58, MACs (G)=925, FPS=21.9, Resolution=512 x 20482025.0349.1
CAFormer-S18Framework=UperNet, Params (M)=54, MACs (G)=1024, FPS=33.0, Resolution=512 x 20482025.0348.9
SPANetV2-S18-pureFramework=UperNet, Params (M)=60, MACs (G)=930, FPS=18.8, Resolution=512 x 20482025.0348.7
ConvFormer-S18Framework=UperNet, Params (M)=54, MACs (G)=925, FPS=36.6, Resolution=512 x 20482025.0348.6
SPANetV2-S36-hybridFramework=Semantic FPN, Params (M)=45, MACs (G)=74, FPS=14.3, Resolution=512 x 5122025.0348.6
Conv2Former-TFramework=UperNet, Params (M)=56, Resolution=512 x 20482025.0348
SPANetV2-S36-pureFramework=Semantic FPN, Params (M)=48, MACs (G)=64, FPS=11.4, Resolution=512 x 5122025.0347.9
SPANetV2-S18-hybridFramework=Semantic FPN, Params (M)=31, MACs (G)=48, FPS=24.7, Resolution=512 x 5122025.0347.8
DFFormer-S36Framework=Semantic FPN, Params (M)=50, MACs (G)=60, FPS=10.5, Resolution=512 x 5122025.0347.5
KAT-BFramework=UperNet, #Param.=142M, FLOPs=2011G, Input size=512 x 20482024.0947.4
DeiT-BFramework=UperNet, #Param.=142M, FLOPs=2007G, Input size=512 x 20482024.0947.2
ConvNeXt-TFramework=UperNet, #Param.=60M, FLOPs=939G, Input size=512 x 20482024.0946.7
ConvNeXt-TFramework=UperNet, Params (M)=60, MACs (G)=939, FPS=36.6, Resolution=512 x 20482025.0346.7
SPANetV2-S18-pureFramework=Semantic FPN, Params (M)=33, MACs (G)=43, FPS=18.0, Resolution=512 x 5122025.0346.7
CDFFormer-S36Framework=Semantic FPN, Params (M)=50, MACs (G)=61, FPS=19.8, Resolution=512 x 5122025.0346.7
SPANet-MFramework=Semantic FPN, Params (M)=45, MACs (G)=57, FPS=8.2, Resolution=512 x 5122025.0346.2
KAT-SFramework=UperNet, #Param.=57M, FLOPs=1219G, Input size=512 x 20482024.0946.1
Swin-TFramework=UperNet, #Param.=60M, FLOPs=945G, Input size=512 x 20482024.0945.8
Swin-TFramework=UperNet, Params (M)=60, MACs (G)=945, FPS=40.8, Resolution=512 x 20482025.0345.8
LITv2-MFramework=Semantic FPN, Params (M)=52, MACs (G)=63, Resolution=512 x 5122025.0345.7
InceptionNeXt-SFramework=Semantic FPN, Params (M)=50, MACs (G)=65, Resolution=512 x 5122025.0345.6
SPANet-SFramework=Semantic FPN, Params (M)=32, MACs (G)=46, FPS=10.6, Resolution=512 x 5122025.0345.4
Swin-SFramework=Semantic FPN, Params (M)=53, MACs (G)=70, FPS=31.1, Resolution=512 x 5122025.0345.2
DFFormer-S18Framework=Semantic FPN, Params (M)=35, MACs (G)=41, FPS=17.0, Resolution=512 x 5122025.0345.1
CDFFormer-S18Framework=Semantic FPN, Params (M)=35, MACs (G)=42, FPS=26.3, Resolution=512 x 5122025.0344.9
LITv2-SFramework=Semantic FPN, Params (M)=31, MACs (G)=41, Resolution=512 x 5122025.0344.3
DeiT-SFramework=UperNet, #Param.=57M, FLOPs=1217G, Input size=512 x 20482024.0943.5
InceptionNeXt-TFramework=Semantic FPN, Params (M)=28, MACs (G)=44, Resolution=512 x 5122025.0343.1
PoolFormer-S36Framework=Semantic FPN, Params (M)=35, MACs (G)=48, FPS=25.7, Resolution=512 x 5122025.0342
Swin-TFramework=Semantic FPN, Params (M)=32, MACs (G)=46, FPS=47.8, Resolution=512 x 5122025.0341.5
PoolFormer-S24Framework=Semantic FPN, Params (M)=25, MACs (G)=39, FPS=34.2, Resolution=512 x 5122025.0340.3
ResNet-101Framework=Semantic FPN, Params (M)=48, MACs (G)=65, FPS=33.5, Resolution=512 x 5122025.0338.8
ResNet-50Framework=Semantic FPN, Params (M)=29, MACs (G)=46, FPS=33.5, Resolution=512 x 5122025.0336.7
MAFT+VLM Backbone=ConvNeXt-L, Ensemble Operation=false2024.0836.1
MAFTVLM Backbone=ConvNeXt-L, Ensemble Operation=false2024.0834.4
FC-CLIPVLM Backbone=ConvNeXt-L, Ensemble Operation=true2024.0834
MAFTVLM Backbone=ViT-L, Ensemble Operation=false2024.0833
SANVLM Backbone=ViT-L, Ensemble Operation=false2024.0832.1
ODISEVLM Backbone=ViT-L, Ensemble Operation=false2024.0829.9
OVSegVLM Backbone=ViT-L, Ensemble Operation=false2024.0829.6
OpenSegVLM Backbone=ALIGN, Ensemble Operation=false2024.0828.6
FC-CLIPVLM Backbone=ConvNeXt-L, Ensemble Operation=false2024.0826.6