A new public coronary-angiography benchmark finds ConvNeXt V2 + DeepLabV3+ is the best single model (macro F1 = 0.456), with a three-model ensemble reaching 0.479.
MedNeXt: Transformer- driven scaling of convnets for medical im- age segmentation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CARDIAG: A Dense Segment Classification Benchmark of Deep Learning Architectures for Coronary Angiography
A new public coronary-angiography benchmark finds ConvNeXt V2 + DeepLabV3+ is the best single model (macro F1 = 0.456), with a three-model ensemble reaching 0.479.