Out Of Distribution Detection On Imagenet 1K 3

评估指标

AUROC

FPR95

Latency, ms

评测结果

各个模型在此基准测试上的表现结果

				Paper Title	Repository
GEN	-	68.32	-	GEN: Pushing the Limits of Softmax-Based Out-of-Distribution Detection	-
KNN (ResNet-50)	86.2	59.08	-	Out-of-Distribution Detection with Deep Nearest Neighbors
DOE	85.98	55.87	-	Out-of-distribution Detection with Implicit Outlier Transformation
GradNorm	-	50.03	-	On the Importance of Gradients for Detecting Distributional Shifts in the Wild
DML	-	47.32	-	Decoupling MaxLogit for Out-of-Distribution Detection	-
RP+GradNorm	-	43.87	-	Detecting Out-of-distribution Data through In-distribution Class Prior	-
ReAct (ResNet-50)	91.53	42.40	-	ReAct: Out-of-distribution Detection With Rectified Activations
RankFeat (ResNetv2-101)	91.91	41.31	-	RankFeat: Rank-1 Feature Removal for Out-of-distribution Detection
SHE	-	34.22	-	Out-of-Distribution Detection based on In-Distribution Data Patterns Memorization with Modern Hopfield Energy	-
MCM (CLIP-B)	94.61	30.91	-	Delving into Out-of-Distribution Detection with Vision-Language Representations
KNN (ResNet-50 SupCon)	94.72	30.83	-	Out-of-Distribution Detection with Deep Nearest Neighbors
MCM (CLIP-L)	94.95	28.38	-	Delving into Out-of-Distribution Detection with Vision-Language Representations
ODIN+UMAP (ResNet-50)	94.71	21.97	-	Unleashing Mask: Explore the Intrinsic Out-of-Distribution Detection Capability
DICE + ReAct (ResNet-50)	96.24	18.64	-	DICE: Leveraging Sparsification for Out-of-Distribution Detection
NPOS	96.19	16.58	-	Non-Parametric Outlier Synthesis
LINe (ResNet-50)	97.56	12.26	-	LINe: Out-of-Distribution Detection by Leveraging Important Neurons
ASH-S (ResNet-50)	97.87	11.49	-	Extremely Simple Activation Shaping for Out-of-Distribution Detection
NNGuide (ResNet50 w/ ReAct)	97.7	11.12	11.10	Nearest Neighbor Guidance for Out-of-Distribution Detection
SCALE (ResNet50)	98.17	9.5	11.27	Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhancement
MOS (BiT-S-R101x1)	98.15	9.28	-	MOS: Towards Scaling Out-of-distribution Detection for Large Semantic Space

0 of 27 row(s) selected.