List your company

Multi-task Learning (Segmentation, Part Segmentation, Disparity) on Cityscapes

70.62Semantic Segmentation mIoUMTAN
54.406458.615762.82567.0343Dec 23, 2025

Evaluation results

MethodLinks
MTANBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.29, #F (Relative FLOPs)=1.482025.1270.62—0.01250.49
TSσBNBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.00, #F (Relative FLOPs)=1.442025.1270.17—0.01230.85
CSBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.42, #F (Relative FLOPs)=1.442025.1269.97—0.01230.55
TSBNBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.00, #F (Relative FLOPs)=1.442025.1269.89—0.01240.38
HPSBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.00, #F (Relative FLOPs)=1.002025.1269.81—0.01250
MMOEBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.42, #F (Relative FLOPs)=1.442025.1269.81—0.0126-0.43
LTBBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.42, #F (Relative FLOPs)=1.442025.1269.81—0.0125-0.35
CGCBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.85, #F (Relative FLOPs)=1.882025.1269.75—0.0125-0.12
DSelect-kBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.44, #F (Relative FLOPs)=1.442025.1269.67—0.01240.26
PLEBackbone=ResNet-50, Architecture Framework=LibMTL / DeepLabV3, #P (Relative Parameters)=1.95, #F (Relative FLOPs)=2.322025.1269.3—0.0129-2.02
CSRelative Parameters (#P)=1.002025.1256.7353.890.7812.43
STLRelative Parameters (#P)=1.002025.1256.6153.950.8410
TSσBNRelative Parameters (#P)=0.602025.1256.4553.260.8140.57
TSBNRelative Parameters (#P)=0.612025.1256.152.820.8060.4
MTANRelative Parameters (#P)=0.782025.1255.8352.610.7990.39
HPSRelative Parameters (#P)=0.602025.1255.0351.920.796-0.39