REVIEW 33 cited by
A ConvNet for the 2020s
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
A ConvNet for the 2020s
read the original abstract
The "Roaring 20s" of visual recognition began with the introduction of Vision Transformers (ViTs), which quickly superseded ConvNets as the state-of-the-art image classification model. A vanilla ViT, on the other hand, faces difficulties when applied to general computer vision tasks such as object detection and semantic segmentation. It is the hierarchical Transformers (e.g., Swin Transformers) that reintroduced several ConvNet priors, making Transformers practically viable as a generic vision backbone and demonstrating remarkable performance on a wide variety of vision tasks. However, the effectiveness of such hybrid approaches is still largely credited to the intrinsic superiority of Transformers, rather than the inherent inductive biases of convolutions. In this work, we reexamine the design spaces and test the limits of what a pure ConvNet can achieve. We gradually "modernize" a standard ResNet toward the design of a vision Transformer, and discover several key components that contribute to the performance difference along the way. The outcome of this exploration is a family of pure ConvNet models dubbed ConvNeXt. Constructed entirely from standard ConvNet modules, ConvNeXts compete favorably with Transformers in terms of accuracy and scalability, achieving 87.8% ImageNet top-1 accuracy and outperforming Swin Transformers on COCO detection and ADE20K segmentation, while maintaining the simplicity and efficiency of standard ConvNets.
Forward citations
Cited by 33 Pith papers
-
AIMIP Phase 1: systematic evaluations of AI weather and climate models
AI atmosphere models trained only on ERA5 reproduce historical climate and ENSO as well as GFDL-CM4, but several under-estimate warming trends and produce implausible responses to uniform +2 K/+4 K SST perturbations.
-
AIMIP Phase 1: systematic evaluations of AI weather and climate models
Under one protocol, most AI climate models reproduce historical climatology and ENSO response as well as a CMIP6 model, but some underestimate warming trends and all diverge on +2/+4K SST experiments.
-
Generative diffusion models for spatiotemporal influenza forecasting
Influpaint uses generative diffusion models on image-encoded influenza data to produce realistic and diverse epidemic trajectories that match leading ensemble methods in accuracy.
-
Descriptor: Parasitoid Wasps and Associated Hymenoptera Dataset (DAPWH)
Releases the DAPWH dataset of 3556 wasp images including 1739 COCO-annotated examples to enable AI models for identifying Ichneumonoidea and associated families.
-
ROMS-IMLE: A Minimalist Approach to Competitive Single-Step Generative Modelling
A single-step IMLE generator with per-stage supervision and a robust loss reports FID 2.56 on ImageNet-256 by filtering ~5% of samples at test time.
-
C3DIR: A Deep Learning 3-Dimensional Cloud Property Retrieval Scheme for Passive Satellite Imagers
C3DIR is a single multi-sensor deep-learning model that retrieves 3-D ice, liquid, and rain water content from passive imagers using voxel-to-voxel collocation with active-sensor profiles.
-
Continuously Evolving Deepfake Detection: An Architecture and Public-Benchmark Evaluation of a Dynamic Detection System
A continuously refreshed, incentive-driven deepfake detector beats static detectors on in-the-wild benchmarks and improves on post-export AI-generated media.
-
ScaleAware-JEPA: Latent Representation for Discovery in Multiscale Physical Fields
ScaleAware-JEPA combines Constrained Diffusion Decomposition with a scale-tied JEPA objective to learn label-free latent coordinates that recover coherent morphology in multiscale fields such as MHD turbulence and int...
-
Modeling Local, Global, and Cross-Modal Context in Multimodal 3D MRI
MICViT outperforms CNN and transformer baselines on brain age prediction from multimodal 3D MRI by combining modality-specific and cross-modal local/global attention across three heterogeneous datasets.
-
TimeProVe: Propose, then Verify for Efficient Long Video Temporal Reasoning in Activities of Daily Living
TimeProVe proposes a propose-then-verify framework using lightweight action-based candidate evidence generation followed by targeted VLM verification for efficient long video temporal reasoning, achieving 7.3% improve...
-
Spiral arms across cosmic time: JWST measurements of the pitch angles of spiral galaxies at $z<3.5$
JWST measurements of pitch angles in 593 spiral galaxies to z=3.5 show no overall redshift evolution but reveal correlations with mass and sSFR only below z=1.25, implying a transition from locally driven to globally ...
-
Generating synthetic computed tomography for radiotherapy: SynthRAD2025 challenge report
SynthRAD2025 shows deep learning produces synthetic CTs with MAE 48-65 HU and high dosimetric gamma passing rates for radiotherapy, performing better on CBCT-to-CT than MRI-to-CT tasks.
-
Birds of a Feather Flock Together: Background-Invariant Representations via Linear Structure in VLMs
Exploiting linear structure in VLM embeddings, a synthetic-data pre-training method yields background-invariant representations that exceed 90% worst-group accuracy on Waterbirds even under 100% spurious correlation w...
-
AIMIP Phase 1: systematic evaluations of AI weather and climate models
AIMIP Phase 1 sets up a common experiment and five evaluation criteria for AI atmosphere models forced by historical sea surface temperatures, finding they match conventional models on most metrics but underestimate s...
-
AIMIP Phase 1: systematic evaluations of AI weather and climate models
AIMIP Phase 1 shows AI models simulate historical climate and El Niño responses as well as traditional models, though some underestimate trends and diverge in generalization tests, with a public dataset released for f...
-
A Novel Graph-Regulated Disentangling Mamba Model with Sparse Tokens for Enhanced Tree Species Classification from MODIS Time Series
A graph-regulated disentangling Mamba model with sparse tokens achieves 93.94% accuracy classifying tree species from MODIS time series in Alberta and outperforms twelve prior models.
-
Only Brains Align with Brains: Cross-Region Alignment Patterns Expose Limits of Normative Models
Alignment pattern analysis reveals that models aligned to individual brain ROIs do not reproduce the stable cross-region alignment profiles observed across human subjects.
-
Towards Real-Time ECG and EMG Modeling on $\mu$NPUs
PhysioLite delivers Transformer-comparable ECG/EMG performance using learnable wavelet filters and hardware-aware design at ~370KB quantized size on μNPUs.
-
Geographically-Weighted Weakly Supervised Bayesian High-Resolution Transformer for 200m Resolution Pan-Arctic Sea Ice Concentration Mapping and Uncertainty Estimation using Sentinel-1, RCM, and AMSR2 Data
A 200m pan-Arctic sea ice concentration model using a Bayesian Transformer with a geographically-weighted weakly supervised loss; reported 0.70 feature detection accuracy and R²=0.90 vs ASI.
-
Model soups need only one ingredient
A single checkpoint, edited by splitting each layer's update with SVD and reweighting the high- and low-energy parts, reaches soup-level OOD robustness without multi-model training.
-
Small, Bias-Free, Blind and Convolutional Denoiser: A compact ConvNeXt U-Net for blind Gaussian color-image denoising
BF-ConvUNeXt, a 0.82M-parameter bias-free ConvNeXt U-Net, is degree-1 homogeneous and matches DnCNN/FFDNet in blind color denoising, extrapolating smoothly beyond its training noise range.
-
Leveraging Multimodality for Real-Time Classification of Transients and Variables found by the Zwicky Transient Facility
ORACLE-2 multimodal classifiers raise macro F1 from 0.52-0.66 (light-curve only) to 0.73 on ZTF Bright Transient Survey data and reach 0.88 on simulated ELAsTiCC data.
-
SPADE: Sketch-guided Path Planning Augmented with Diffusion Experts
SPADE combines sketch-guided path planning with diffusion-augmented imitation learning to achieve better generalization and lower error with fewer parameters than prior methods.
-
Hist2Style: Histogram-Guided Stylization with Bilateral Grids
Hist2Style introduces a lightweight bilateral-grid network conditioned on histogram embeddings for distilling large-model stylization into real-time, structure-preserving, user-controllable photorealistic edits.
-
Physics-informed convolutional neural networks for fluid flow through porous media
A physics-informed CNN predicts pore-scale velocity fields from geometry and serves as a warm-start to accelerate Lattice-Boltzmann solvers in over 90% of tested cases.
-
Federated Medical Image Classification under Class and Domain Imbalance exploiting Synthetic Sample Generation
FedSSG generates and shares synthetic samples within a federated setup to reduce class imbalance and domain shift problems in medical image classification.
-
HLGFA: High-Low Resolution Guided Feature Alignment for Unsupervised Anomaly Detection
HLGFA detects anomalies by identifying breakdowns in cross-resolution feature consistency between high- and low-resolution views of normal samples, guided by structure and detail priors, and reports 97.9% pixel AUROC ...
-
InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks
InternVL scales a vision model to 6B parameters and aligns it with LLMs using web data to achieve state-of-the-art results on 32 visual-linguistic benchmarks.
-
A Controlled Visual-Backbone Benchmark for Multimodal Short-Term Solar Irradiance Forecasting
Under one fixed multimodal 10-min irradiance forecaster, VMamba-S and Swin-B nearly tie best Folsom RMSE (~65.4 W/m²) over smart persistence, while NREL’s tiny matched split favors persistence.
-
Multi-Task Learning for Heterogeneous Prediction from Video Game State with Transfer Learning
On a large World of Tanks dataset, a shared multi-task model with equal weighting or PCGrad outperforms single-task models on average, and task/map pre-training helps most in low-data regimes.
-
LETT-NeXt: A Lightweight RECIST-Guided Model for 3D CT Lesion Segmentation
LETT-NeXt uses RECIST line prompts in a cropped MedNeXt-v2 encoder-decoder to predict 3D lesion masks, reaching DSC 73.9 on hidden test data for a CVPR 2026 segmentation competition.
-
NTIRE 2026 The Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results
The NTIRE 2026 challenge reports strong performance from 17 teams on raindrop removal for dual-focused day and night images using an adjusted real-world dataset with 14,139 training images.
-
NTIRE 2026 The Second Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results
The second NTIRE challenge on day and night raindrop removal for dual-focused images received 17 valid team submissions that demonstrated strong performance on the Raindrop Clarity dataset.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.