Papers › Architecture-Agnostic Masked Image Modeling -- From ViT back to CNN

Architecture-Agnostic Masked Image Modeling -- From ViT back to CNN

27 May 2022arXiv:2205.13943archive 2025-07-28

Siyuan Li, Di wu, Fang Wu, Zelin Zang, Stan. Z. Li

Masked image modeling, an emerging self-supervised pre-training method, has shown impressive success across numerous downstream vision tasks with Vision transformers. Its underlying idea is simple: a portion of the input image is masked out and then reconstructed via a pre-text task. However, the working principle behind MIM is not well explained, and previous studies insist that MIM primarily works for the Transformer family but is incompatible with CNNs. In this work, we observe that MIM essentially teaches the model to learn better middle-order interactions among patches for more generalized feature extraction. We then propose an Architecture-Agnostic Masked Image Modeling framework (A²MIM), which is compatible with both Transformers and CNNs in a unified way. Extensive experiments on popular benchmarks show that A²MIM learns better representations without explicit design and endows the backbone model with the stronger capability to transfer to various downstream tasks.

PaperPDFCode

In Syntology View this paper on Syntology, its page in Syntology's graph. That page lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Westlake-AI/A2MIM officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: not harvested report
Westlake-AI/openmixup officialmentioned in papermentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to mixup: Beyond Empirical Risk Minimization (arXiv:1710.09412; that paper's own run record: 36 ran (of which 5 constructed an object rather than computing a result; 14 with no instrument failure: 0 honoured, 0 violated, 14 with no contract checked; 22 where Syntology's instrument failed) · 11 unverified), Visualizing the Loss Landscape of Neural Nets (arXiv:1712.09913; that paper's own run record: 3 ran (of which 1 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 1 where Syntology's instrument failed) · 5 unverified), CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features (arXiv:1905.04899; that paper's own run record: 17 ran (of which 6 constructed an object rather than computing a result; 10 with no instrument failure: 2 honoured, 0 violated, 8 with no contract checked; 7 where Syntology's instrument failed) · 7 unverified) and 1 more (every one in this page's JSON twin); nothing checks that they implement this paper's method. report
open-mmlab/mmpretrain pytorchApache-2.0Syntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Improving Pixel-based MIM by Reducing Wasted Modeling Capability (arXiv:2308.00261; that paper's own run record: 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationInstance SegmentationObject DetectionSelf-Supervised Image ClassificationSelf-Supervised LearningSemantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Instance Segmentation COCO test-dev A2MIM (ViT-B) mask AP 43.5 #48 of 112 Archive leaderboard report
Instance Segmentation COCO test-dev A2MIM (ResNet-50 2x) mask AP 34.9 #98 of 112 Archive leaderboard report
Object Detection COCO test-dev A2MIM (ViT-B) box mAP 49.4 #96 of 225 Archive leaderboard report
Object Detection COCO test-dev A2MIM (ResNet-50 2x) box mAP 39.8 #202 of 225 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM+ (ViT-B) Top 1 Accuracy 84.5% #30 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM (ViT-B) Top 1 Accuracy 84.2% #34 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM+ (ViT-S) Top 1 Accuracy 82.4% #51 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM (ViT-S) Top 1 Accuracy 82.2% #53 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM+ (ResNet-50 RSB-A2) Top 1 Accuracy 80.5% #56 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM (ResNet-50 RSB-A2) Top 1 Accuracy 80.4% #57 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM+ (ResNet-50 RSB-A3) Top 1 Accuracy 78.9% #58 of 65 Archive leaderboard report
Self-Supervised Image Classification ImageNet (finetuned) A2MIM (ResNet-50 RSB-A3) Top 1 Accuracy 78.8% #59 of 65 Archive leaderboard report
Semantic Segmentation ADE20K A2MIM (ViT-B) Validation mIoU 49 #141 of 235 Archive leaderboard report
Semantic Segmentation ADE20K A2MIM (ResNet-50) Validation mIoU 38.3 #222 of 235 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMIMMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections