Thermal Image Segmentation On Pst900

评估指标

mIoU

评测结果

各个模型在此基准测试上的表现结果

模型名称	mIoU	Paper Title	Repository
CAINet (MobileNet-V2)	84.74	Context-Aware Interaction Network for RGB-T Semantic Segmentation
PSTNet	68.4	PST900: RGB-Thermal Calibration, Dataset and Segmentation Network
Fast-SCNN	47.2	Fast-SCNN: Fast Semantic Segmentation Network
SHIFNet	89.8	Unveiling the Potential of Segment Anything Model 2 for RGB-Thermal Semantic Segmentation with Language Guidance
UniRGB-IR	82.8	UniRGB-IR: A Unified Framework for RGB-Infrared Semantic Tasks via Adapter Tuning
RTFNet	57.6	RTFNet: RGB-Thermal Fusion Network for Semantic Segmentation of Urban Scenes
CSRPNet (ResNet-50)	75.0	Channel and Spatial Relation-Propagation Network for RGB-Thermal Semantic Segmentation	-
EGFNet	78.51	Edge-aware Guidance Fusion Network for RGB Thermal Scene Parsing
MTANet	78.60	MTANet: Multitask-Aware Network With Hierarchical Multimodal Fusion for RGB-T Urban Scene Understanding	-
CACFNet	86.56	CACFNet: Cross-Modal Attention Cascaded Fusion Network for RGB-T Urban Scene Parsing	-
EGFNet (ConvNeXt)	85.42	EGFNet: Edge-Aware Guidance Fusion Network for RGB–Thermal Urban Scene Parsing
CRM_RGBTSeg	88	Complementary Random Masking for RGB-Thermal Semantic Segmentation
MMSFormer	87.45	MMSFormer: Multimodal Transformer for Material and Semantic Segmentation
Sigma-small	87.8	Sigma: Siamese Mamba Network for Multi-Modal Semantic Segmentation
HAPNet	89.0	HAPNet: Toward Superior RGB-Thermal Scene Parsing via Hybrid, Asymmetric, and Progressive Heterogeneous Feature Fusion
GEBNet	81.15	GEBNet: Graph-Enhancement Branch Network for RGB-T Scene Parsing	-
MFNet	57.0	MFNet: Towards real-time semantic segmentation for autonomous vehicles with multi-spectral scenes	-
ACNet	71.81	ACNet: Attention Based Network to Exploit Complementary Features for RGBD Semantic Segmentation
SpiderMesh	82.3	SpiderMesh: Spatial-aware Demand-guided Recursive Meshing for RGB-T Semantic Segmentation
DPLNet	86.7	Efficient Multimodal Semantic Segmentation via Dual-Prompt Learning

0 of 22 row(s) selected.