Audio Classification On Icbhi Respiratory

评估指标

ICBHI Score

Sensitivity

Specificity

评测结果

各个模型在此基准测试上的表现结果

模型名称	ICBHI Score	Sensitivity	Specificity	Paper Title	Repository
DAT (AST)	59.81	42.50	77.11	Stethoscope-guided Supervised Contrastive Learning for Cross-domain Adaptation on Respiratory Sound Classification
AST (Patch-Mix CL)	62.37	43.07	81.66	Patch-Mix Contrastive Learning with Audio Spectrogram Transformer on Respiratory Sound Classification	-
SG-SCL (AST)	61.71	43.55	79.87	Stethoscope-guided Supervised Contrastive Learning for Cross-domain Adaptation on Respiratory Sound Classification
CNN6 (+metadata)	58.04	-	-	Pretraining Respiratory Sound Representations using Metadata and Contrastive Learning
AST (fine-tuning)	-	41.97	77.14	Patch-Mix Contrastive Learning with Audio Spectrogram Transformer on Respiratory Sound Classification	-
ResNeSt	55.30	40.20	70.40	A DOMAIN TRANSFER BASED DATA AUGMENTATION METHOD FOR AUTOMATED RESPIRATORY CLASSIFICATION	-
Audio-CLAP	62.56	44.67	80.85	BTS: Bridging Text and Sound Modalities for Metadata-Aided Respiratory Sound Classification
bi-ResNet (scratch)	50.16	31.10	69.20	LungBRN: A Smart Digital Stethoscope for Detecting Respiratory Disease Using bi-ResNet Deep Learning Algorithm
M2D/0.7 (e=0.3)	62.73	-	-	Masked Modeling Duo: Towards a Universal Audio Pre-training Framework
CNN6 (scratch)	54.74	33.84	75.35	Pretraining Respiratory Sound Representations using Metadata and Contrastive Learning
CycleGuardian	63.26	44.47	82.06	CycleGuardian: A Framework for Automatic RespiratorySound classification Based on Improved Deep clustering and Contrastive Learning
AFT on Mixed-500	61.79	42.86	80.72	Adversarial Fine-tuning using Generated Respiratory Sound to Address Class Imbalance
ResNet-34	56.20	40.10	72.30	RespireNet: A Deep Neural Network for Accurately Detecting Abnormal Lung Sounds in Limited Data Setting
CNN6	57.55	-	75.95	Pretraining Respiratory Sound Representations using Metadata and Contrastive Learning
CNN6 (+metadata)	-	39.15	76.93	Pretraining Respiratory Sound Representations using Metadata and Contrastive Learning
M2D-X/0.7 (η=0.3)	63.29	-	-	0/1 Deep Neural Networks via Block Coordinate Descent	-
ResNet-50	58.29	37.24	79.34	Lung Sound Classification Using Co-tuning and Stochastic Normalization	-
AST (fine-tuning)	59.55	-	-	Patch-Mix Contrastive Learning with Audio Spectrogram Transformer on Respiratory Sound Classification	-
BTS	63.54	45.67	81.4	BTS: Bridging Text and Sound Modalities for Metadata-Aided Respiratory Sound Classification
bi-ResNet-Att	56.76	46.38	67.13	ARSC-Net: Adventitious Respiratory Sound Classification Network Using Parallel Paths with Channel-Spatial Attention	-

0 of 22 row(s) selected.