Multi Label Classification On Ms Coco

评估指标

mAP

评测结果

各个模型在此基准测试上的表现结果

模型名称	mAP	Paper Title	Repository
ResNet-SRN	77.1	Learning Spatial Regularization with Image-level Supervisions for Multi-label Image Classification
MlTr-XL(ImageNet-21K pretraining, resolution 384)	90.0	MlTr: Multi-label Classification with Transformer
MCAR (ResNet101, 448x448)	83.8	Learning to Discover Multi-Class Attentional Regions for Multi-Label Image Recognition
MS-CMA	83.8	Cross-Modality Attention with Semantic Graph Embedding for Multi-Label Classification	-
ADD-GCN	85.2	Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition
MCAR (ResNet101, 576x576)	84.5	Learning to Discover Multi-Class Attentional Regions for Multi-Label Image Recognition
Q2L-CvT(ImageNet-21K pretraining, resolution 384)	91.3	Query2Label: A Simple Transformer Way to Multi-Label Classification
Q2L-R101(resolution 448)	84.9	Query2Label: A Simple Transformer Way to Multi-Label Classification
GKGNet(resolution 576)	87.7	GKGNet: Group K-Nearest Neighbor based Graph Convolutional Network for Multi-Label Image Recognition
ML-Decoder(TResNet-XL, resolution 640)	91.4	ML-Decoder: Scalable and Versatile Classification Head
MlTr-L(ImageNet-21K pretraining, resolution 384)	88.5	MlTr: Multi-label Classification with Transformer
Q2L-SwinL(ImageNet-21K pretraining, resolution 384)	90.5	Query2Label: A Simple Transformer Way to Multi-Label Classification
MLD-TResNet-L-AAM[640x640]	91.30	Combining Metric Learning and Attention Heads For Accurate and Efficient Multilabel Image Classification
TResNet-XL (resolution 640)	88.4	Asymmetric Loss For Multi-Label Classification
IDA-SwinL	90.3	Causality Compensated Attention for Contextual Biased Visual Recognition
CCD-SwinL	90.3	Contextual Debiasing for Visual Recognition With Causal Mechanisms
ADDS(ViT-L-336, resolution 640)	93.41	Open Vocabulary Multi-Label Classification with Dual-Modal Decoder on Aligned Visual-Textual Features	-
MSRN	83.4	Multi-layered Semantic Representation Network for Multi-label Image Classification
TResNet-L (resolution 448)	86.6	Asymmetric Loss For Multi-Label Classification
KSSNet	83.7	Multi-Label Classification with Label Graph Superimposing

0 of 34 row(s) selected.