HyperAI

Image Classification On Mini Webvision 1 0

Métriques

ImageNet Top-1 Accuracy
ImageNet Top-5 Accuracy
Top-1 Accuracy
Top-5 Accuracy

Résultats

Résultats de performance de divers modèles sur ce benchmark

Nom du modèle
ImageNet Top-1 Accuracy
ImageNet Top-5 Accuracy
Top-1 Accuracy
Top-5 Accuracy
Paper TitleRepository
Co-teaching (Inception-ResNet-v2)61.4884.7063.5885.20Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels
DivideMix with C2D (ResNet-50)78.57 ± 0.3793.04 ± 0.1079.42 ± 0.3492.32 ± 0.33Contrast to Divide: Self-Supervised Pre-Training for Learning with Noisy Labels
PSSCL (130 epochs)79.6895.1679.5694.84PSSCL: A progressive sample selection framework with contrastive loss designed for noisy labels
RTE (Inception-ResNet-v2)80.8497.24--Robust Temporal Ensembling for Learning with Noisy Labels-
Dynamic Loss (Inception-ResNet-v2)74.7693.0880.1293.64Dynamic Loss For Robust Learning
LongReMix (Inception-ResNet-v2)--78.9292.32LongReMix: Robust Learning with High Confidence Samples in a Noisy Label Environment
LRA-diffusion (CLIP ViT)82.56-84.16-Label-Retrieval-Augmented Diffusion Models for Learning from Noisy Labels
NCT (Inception-ResNet-v2)71.7391.6175.1690.77Noisy Concurrent Training for Efficient Learning under Label Noise
ROLT+ (Inception-ResNet-v2)74.6492.4877.6492.44Robust Long-Tailed Learning under Label Noise-
D2L (Inception-ResNet-v2)57.8081.3662.6884.00Dimensionality-Driven Learning with Noisy Labels
CC76.0893.8679.3693.64Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels
PGDF (Inception-ResNet-v2)75.4593.1181.4794.03Sample Prior Guided Robust Model Learning to Suppress Noisy Labels
PSSCL (120 epochs)79.4094.8478.5293.80PSSCL: A progressive sample selection framework with contrastive loss designed for noisy labels
DivideMix (ResNet-50)74.42 ±0.2991.21 ±0.1276.32 ±0.3690.65 ±0.16DivideMix: Learning with Noisy Labels as Semi-supervised Learning
F-Correction (Inception-ResNet-v2)57.3682.3661.1282.68Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach
Crust (Inception-ResNet-v2)67.3687.8472.4089.56Coresets for Robust Training of Neural Networks against Noisy Labels-
TCL75.492.479.192.3Twin Contrastive Learning with Noisy Labels
MOIT+ (ResNet-18)--78.76-Multi-Objective Interpolation Training for Robustness to Label Noise
FaMUS7792.7679.492.80Faster Meta Update Strategy for Noise-Robust Deep Learning
ODD (Inception-ResNet-v2)66.786.374.690.6Robust and On-the-fly Dataset Denoising for Image Classification-
0 of 47 row(s) selected.