Papers › Making Convolutional Networks Shift-Invariant Again

Making Convolutional Networks Shift-Invariant Again

25 Apr 2019arXiv:1904.11486archive 2025-07-28

Richard Zhang

Modern convolutional networks are not shift-invariant, as small input shifts or translations can cause drastic changes in the output. Commonly used downsampling methods, such as max-pooling, strided-convolution, and average-pooling, ignore the sampling theorem. The well-known signal processing fix is anti-aliasing by low-pass filtering before downsampling. However, simply inserting this module into deep networks degrades performance; as a result, it is seldomly used today. We show that when integrated correctly, it is compatible with existing architectural components, such as max-pooling and strided-convolution. We observe increased accuracy in ImageNet classification, across several commonly-used architectures, such as ResNet, DenseNet, and MobileNet, indicating effective regularization. Furthermore, we observe better generalization, in terms of stability and robustness to input corruptions. Our results demonstrate that this classical signal processing technique has been undeservingly overlooked in modern deep networks. Code and anti-aliased versions of popular networks are available at https://richzhang.github.io/antialiased-cnns/ .

PaperPDFConference PDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also 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

Syntology 4 code samples: 2 from 2 repositories linked to this paper, 2 identical to code harvested from another paper's repository. Across all 4: 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code.

By repository: official repository: 1 sample from 1 repository, 0 ran (0 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 1 sample from 1 repository, 0 ran (0 with no instrument failure; 0 where Syntology's instrument failed); 2 identical to code first harvested elsewhere. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 2 of its repositories that Syntology harvested for other papers:

No date is recorded for any of the 2 samples that ran: their levels were written before Syntology's graph kept the date.

adobe/antialiased-cnns officialmentioned on GitHubpytorchSyntology: 1 sample, 0 ran · 1 unverified report
mauriceweiler/MobiusCNNs mentioned on GitHubpytorchGPL-3.0Syntology: not harvested report
mnikitin/Shift-Invariant-CNNs mentioned on GitHubmxnetSyntology: not harvested report
ricky40403/BlurPool mentioned on GitHubSyntology: not harvested report
rwightman/pytorch-image-models mentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to AdamP: Slowing Down the Slowdown for Momentum Optimizers on Scale-invariant Weights (arXiv:2006.08217; that paper's own run record: 4 ran (of which 1 constructed an object rather than computing a result; 2 with no instrument failure: 1 honoured, 0 violated, 1 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
tattaka/Antialiased-CNNs-Converter-PyTorch mentioned on GitHubpytorchSyntology: 1 sample, 0 ran · 1 unverified report
kornia/kornia pytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer (arXiv:2110.02178; that paper's own run record: 53 ran (of which 31 constructed an object rather than computing a result; 44 with no instrument failure: 1 honoured, 0 violated, 43 with no contract checked; 9 where Syntology's instrument failed) · 15 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

4 samples harvested: 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · our draft was wrong
2unverified

Licence: 3 of the 4 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. 2 samples are listed last, in a group without a repository. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

adobe/antialiased-cnns official repository 1 sample, 0 ran · 1 unverified
resnet18 in 432 papers antialiased_cnns/resnet.py unverified licence not identified · pointer only · 27ab42311803bf0e · report
tattaka/Antialiased-CNNs-Converter-PyTorch community (archive-listed) 1 sample, 0 ran · 1 unverified
convert_model in 14 papers antialiased_cnns_converter/antialiased_cnns_converter.py unverified MIT (permissive) · 32c38f429da97bdd · report
Repository not located 2 samples, 2 ran (2 instrument-failed) · 0 unverified
2 samples are identical to code Syntology harvested from another paper's repository; this paper's own copy was not located, so no repository is named and no licence is recorded.
conv1x1 in 691 papers repository not located ran · our draft was wrong licence of this copy not recorded · d9def42110729a85 · report fixture: signature in_planes:int/0/int, out_planes:int/0/int, stride:int/0/int · returned Conv2d · level recorded, date not recorded
conv3x3 in 1,376 papers repository not located ran · our draft was wrong licence of this copy not recorded · 160bb14bd76201b4 · report fixture: signature in_planes:int/0/int, out_planes:int/0/int, stride:int/0/int, groups:int/0/int, dilation:int/0/int · returned Conv2d · level recorded, date not recorded

Tasks

Classification ConsistencyConditional Image GenerationDomain GeneralizationImage ClassificationImage Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Domain Generalization VizWiz-Classification ResNet-101 (lpf3) Accuracy - All Images 41.7 #26 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf3) Accuracy - Clean Images 46.1 #26 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf3) Accuracy - Corrupted Images 35.7 #26 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf5) Accuracy - All Images 41.5 #29 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf5) Accuracy - Clean Images 45.3 #29 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf5) Accuracy - Corrupted Images 35.2 #29 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf2) Accuracy - All Images 41.1 #31 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf2) Accuracy - Clean Images 45.2 #31 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf2) Accuracy - Corrupted Images 35.1 #31 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf5) Accuracy - All Images 41 #32 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf5) Accuracy - Clean Images 45.8 #32 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-101 (lpf5) Accuracy - Corrupted Images 34.8 #32 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf2) Accuracy - All Images 40.3 #36 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf2) Accuracy - Clean Images 45.1 #36 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf2) Accuracy - Corrupted Images 33.4 #36 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf3) Accuracy - All Images 40 #37 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf3) Accuracy - Clean Images 44.7 #37 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-50 (lpf3) Accuracy - Corrupted Images 34.3 #37 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet121 (lpf5) Accuracy - All Images 38.7 #43 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet121 (lpf5) Accuracy - Clean Images 42.7 #43 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet121 (lpf5) Accuracy - Corrupted Images 32 #43 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf2) Accuracy - All Images 38.3 #44 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf2) Accuracy - Clean Images 42.8 #44 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf2) Accuracy - Corrupted Images 32.4 #44 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf3) Accuracy - All Images 38.3 #45 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf3) Accuracy - Clean Images 42.8 #45 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf3) Accuracy - Corrupted Images 32.3 #45 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf3) Accuracy - All Images 38.3 #46 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf3) Accuracy - Clean Images 42.9 #46 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf3) Accuracy - Corrupted Images 31.9 #46 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf2) Accuracy - All Images 38.3 #47 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf2) Accuracy - Clean Images 43.1 #47 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification DenseNet-121 (lpf2) Accuracy - Corrupted Images 31.7 #47 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf2) Accuracy - All Images 37.2 #53 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf2) Accuracy - Clean Images 41.8 #53 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf2) Accuracy - Corrupted Images 31.3 #53 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf5) Accuracy - All Images 37.2 #54 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf5) Accuracy - Clean Images 42.5 #54 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-34 (lpf5) Accuracy - Corrupted Images 29.9 #54 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf5) Accuracy - All Images 37 #55 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf5) Accuracy - Clean Images 41.7 #55 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf5) Accuracy - Corrupted Images 30.8 #55 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf3) Accuracy - All Images 36.9 #56 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf3) Accuracy - Clean Images 42.1 #56 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 BN (lpf3) Accuracy - Corrupted Images 30.6 #56 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf3) Accuracy - All Images 36 #64 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf3) Accuracy - Clean Images 40.3 #64 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf3) Accuracy - Corrupted Images 30.4 #64 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf5) Accuracy - All Images 35.8 #66 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf5) Accuracy - Clean Images 40.1 #66 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf5) Accuracy - Corrupted Images 29.1 #66 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf3) Accuracy - All Images 35.6 #68 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf3) Accuracy - Clean Images 39.5 #68 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf3) Accuracy - Corrupted Images 28.5 #68 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf2) Accuracy - All Images 35.5 #69 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf2) Accuracy - Clean Images 39.2 #69 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification MobileNetV2 (lpf2) Accuracy - Corrupted Images 30.3 #69 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf2) Accuracy - All Images 35.5 #70 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf2) Accuracy - Clean Images 40.1 #70 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf2) Accuracy - Corrupted Images 28.7 #70 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf3) Accuracy - All Images 35.1 #71 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf3) Accuracy - Clean Images 40 #71 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf3) Accuracy - Corrupted Images 28.2 #71 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf5) Accuracy - All Images 34.7 #75 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf5) Accuracy - Clean Images 38.9 #75 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification ResNet-18 (lpf5) Accuracy - Corrupted Images 27.7 #75 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf5) Accuracy - All Images 34.5 #76 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf5) Accuracy - Clean Images 39.4 #76 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf5) Accuracy - Corrupted Images 27.8 #76 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf2) Accuracy - All Images 33.5 #79 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf2) Accuracy - Clean Images 38.5 #79 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification VGG-16 (lpf2) Accuracy - Corrupted Images 26.7 #79 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf3) Accuracy - All Images 23.1 #87 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf3) Accuracy - Clean Images 26.8 #87 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf3) Accuracy - Corrupted Images 17.5 #87 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf2) Accuracy - All Images 22.8 #88 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf2) Accuracy - Clean Images 26.8 #88 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf2) Accuracy - Corrupted Images 18.2 #88 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf5) Accuracy - All Images 22.7 #89 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf5) Accuracy - Clean Images 26.8 #89 of 90 Archive leaderboard report
Domain Generalization VizWiz-Classification AlexNet (lpf5) Accuracy - Corrupted Images 18.4 #89 of 90 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

Introduced by this paper: Anti-Alias Downsampling

1x1 ConvolutionAnti-Alias DownsamplingAverage PoolingBatch NormalizationBottleneck Residual BlockConcatenated Skip ConnectionConvolutionDense BlockDense ConnectionsDropoutGlobal Average PoolingKaiming InitializationMax PoolingReLUResidual BlockResidual ConnectionSoftmax

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