{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/190411486","title":"Making Convolutional Networks Shift-Invariant Again","arxiv_id":"1904.11486","date":"2019-04-25","proceeding":null,"authors":["Richard Zhang"],"abstract":"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 \\textit{increased accuracy} in ImageNet classification, across several commonly-used architectures, such as ResNet, DenseNet, and MobileNet, indicating effective regularization. Furthermore, we observe \\textit{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/ .","url_abs":"https://arxiv.org/abs/1904.11486v2","url_pdf":"https://arxiv.org/pdf/1904.11486v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"190411486","repo_url":"https://github.com/adobe/antialiased-cnns","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"190411486","repo_url":"https://github.com/mauriceweiler/MobiusCNNs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"190411486","repo_url":"https://github.com/mnikitin/Shift-Invariant-CNNs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"mxnet","reach":{"status":"ok"}},{"paper_slug":"190411486","repo_url":"https://github.com/ricky40403/BlurPool","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"190411486","repo_url":"https://github.com/rwightman/pytorch-image-models","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"190411486","repo_url":"https://github.com/tattaka/Antialiased-CNNs-Converter-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"190411486","repo_url":"https://github.com/kornia/kornia/blob/e4a54c2d4fd9b8eaa88dec50061e33f7aee5d959/kornia/filters/blur_pool.py","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-consistency","task_name":"Classification Consistency"},{"task_slug":"conditional-image-generation","task_name":"Conditional Image Generation"},{"task_slug":"domain-generalization","task_name":"Domain Generalization"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-generation","task_name":"Image Generation"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"anti-alias-downsampling","method_name":"Anti-Alias Downsampling"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"bottleneck-residual-block","method_name":"Bottleneck Residual Block"},{"method_slug":"concatenated-skip-connection","method_name":"Concatenated Skip Connection"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dense-block","method_name":"Dense Block"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"residual-block","method_name":"Residual Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[{"slug":"anti-alias-downsampling","name":"Anti-Alias Downsampling","full_name":"Anti-Alias Downsampling"}],"results":[{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-101 (lpf3)","rank_in_archive_order":26,"of":90,"metrics":{"Accuracy - All Images":"41.7","Accuracy - Clean Images":"46.1","Accuracy - Corrupted Images":"35.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-50 (lpf5)","rank_in_archive_order":29,"of":90,"metrics":{"Accuracy - All Images":"41.5","Accuracy - Clean Images":"45.3","Accuracy - Corrupted Images":"35.2"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-101 (lpf2)","rank_in_archive_order":31,"of":90,"metrics":{"Accuracy - All Images":"41.1","Accuracy - Clean Images":"45.2","Accuracy - Corrupted Images":"35.1"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-101 (lpf5)","rank_in_archive_order":32,"of":90,"metrics":{"Accuracy - All Images":"41","Accuracy - Clean Images":"45.8","Accuracy - Corrupted Images":"34.8"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-50 (lpf2)","rank_in_archive_order":36,"of":90,"metrics":{"Accuracy - All Images":"40.3","Accuracy - Clean Images":"45.1","Accuracy - Corrupted Images":"33.4"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-50 (lpf3)","rank_in_archive_order":37,"of":90,"metrics":{"Accuracy - All Images":"40","Accuracy - Clean Images":"44.7","Accuracy - Corrupted Images":"34.3"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"DenseNet121 (lpf5)","rank_in_archive_order":43,"of":90,"metrics":{"Accuracy - All Images":"38.7","Accuracy - Clean Images":"42.7","Accuracy - Corrupted Images":"32"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-34 (lpf2)","rank_in_archive_order":44,"of":90,"metrics":{"Accuracy - All Images":"38.3","Accuracy - Clean Images":"42.8","Accuracy - Corrupted Images":"32.4"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"DenseNet-121 (lpf3)","rank_in_archive_order":45,"of":90,"metrics":{"Accuracy - All Images":"38.3","Accuracy - Clean Images":"42.8","Accuracy - Corrupted Images":"32.3"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-34 (lpf3)","rank_in_archive_order":46,"of":90,"metrics":{"Accuracy - All Images":"38.3","Accuracy - Clean Images":"42.9","Accuracy - Corrupted Images":"31.9"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"DenseNet-121 (lpf2)","rank_in_archive_order":47,"of":90,"metrics":{"Accuracy - All Images":"38.3","Accuracy - Clean Images":"43.1","Accuracy - Corrupted Images":"31.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 BN (lpf2)","rank_in_archive_order":53,"of":90,"metrics":{"Accuracy - All Images":"37.2","Accuracy - Clean Images":"41.8","Accuracy - Corrupted Images":"31.3"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-34 (lpf5)","rank_in_archive_order":54,"of":90,"metrics":{"Accuracy - All Images":"37.2","Accuracy - Clean Images":"42.5","Accuracy - Corrupted Images":"29.9"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 BN (lpf5)","rank_in_archive_order":55,"of":90,"metrics":{"Accuracy - All Images":"37","Accuracy - Clean Images":"41.7","Accuracy - Corrupted Images":"30.8"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 BN (lpf3)","rank_in_archive_order":56,"of":90,"metrics":{"Accuracy - All Images":"36.9","Accuracy - Clean Images":"42.1","Accuracy - Corrupted Images":"30.6"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"MobileNetV2 (lpf3)","rank_in_archive_order":64,"of":90,"metrics":{"Accuracy - All Images":"36","Accuracy - Clean Images":"40.3","Accuracy - Corrupted Images":"30.4"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"MobileNetV2 (lpf5)","rank_in_archive_order":66,"of":90,"metrics":{"Accuracy - All Images":"35.8","Accuracy - Clean Images":"40.1","Accuracy - Corrupted Images":"29.1"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-18 (lpf3)","rank_in_archive_order":68,"of":90,"metrics":{"Accuracy - All Images":"35.6","Accuracy - Clean Images":"39.5","Accuracy - Corrupted Images":"28.5"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"MobileNetV2 (lpf2)","rank_in_archive_order":69,"of":90,"metrics":{"Accuracy - All Images":"35.5","Accuracy - Clean Images":"39.2","Accuracy - Corrupted Images":"30.3"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-18 (lpf2)","rank_in_archive_order":70,"of":90,"metrics":{"Accuracy - All Images":"35.5","Accuracy - Clean Images":"40.1","Accuracy - Corrupted Images":"28.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 (lpf3)","rank_in_archive_order":71,"of":90,"metrics":{"Accuracy - All Images":"35.1","Accuracy - Clean Images":"40","Accuracy - Corrupted Images":"28.2"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"ResNet-18 (lpf5)","rank_in_archive_order":75,"of":90,"metrics":{"Accuracy - All Images":"34.7","Accuracy - Clean Images":"38.9","Accuracy - Corrupted Images":"27.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 (lpf5)","rank_in_archive_order":76,"of":90,"metrics":{"Accuracy - All Images":"34.5","Accuracy - Clean Images":"39.4","Accuracy - Corrupted Images":"27.8"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"VGG-16 (lpf2)","rank_in_archive_order":79,"of":90,"metrics":{"Accuracy - All Images":"33.5","Accuracy - Clean Images":"38.5","Accuracy - Corrupted Images":"26.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"AlexNet (lpf3)","rank_in_archive_order":87,"of":90,"metrics":{"Accuracy - All Images":"23.1","Accuracy - Clean Images":"26.8","Accuracy - Corrupted Images":"17.5"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"AlexNet (lpf2)","rank_in_archive_order":88,"of":90,"metrics":{"Accuracy - All Images":"22.8","Accuracy - Clean Images":"26.8","Accuracy - Corrupted Images":"18.2"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-vizwiz","task":"Domain Generalization","dataset":"VizWiz-Classification","model":"AlexNet (lpf5)","rank_in_archive_order":89,"of":90,"metrics":{"Accuracy - All Images":"22.7","Accuracy - Clean Images":"26.8","Accuracy - Corrupted Images":"18.4"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.11486","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1904.11486"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/rwightman/pytorch-image-models","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/tattaka/Antialiased-CNNs-Converter-PyTorch","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ricky40403/BlurPool","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/mauriceweiler/MobiusCNNs","reach":{"status":"ok","spdx":"GPL-3.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/adobe/antialiased-cnns","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kornia/kornia/blob/e4a54c2d4fd9b8eaa88dec50061e33f7aee5d959/kornia/filters/blur_pool.py","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/mnikitin/Shift-Invariant-CNNs","reach":{"status":"ok"}}],"summary":{"ran_draft_wrong":2,"unverified":2},"by_repo_kind":{"official":{"samples":1,"ran":0,"repositories":1},"listed":{"samples":1,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"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"},"n_pointer_only_for_licence":3,"samples":[{"code_sha256_prefix":"d9def42110729a85","entry":"conv1x1","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"d9def42110729a85"}},{"code_sha256_prefix":"160bb14bd76201b4","entry":"conv3x3","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"160bb14bd76201b4"}},{"code_sha256_prefix":"32c38f429da97bdd","entry":"convert_model","repo":"tattaka/Antialiased-CNNs-Converter-PyTorch","repo_kind":"listed","path":"antialiased_cnns_converter/antialiased_cnns_converter.py","file_url":"https://github.com/tattaka/Antialiased-CNNs-Converter-PyTorch/blob/HEAD/antialiased_cnns_converter/antialiased_cnns_converter.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"32c38f429da97bdd"}},{"code_sha256_prefix":"27ab42311803bf0e","entry":"resnet18","repo":"adobe/antialiased-cnns","repo_kind":"official","path":"antialiased_cnns/resnet.py","file_url":"https://github.com/adobe/antialiased-cnns/blob/HEAD/antialiased_cnns/resnet.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NOASSERTION","inline_ok":false,"mcp_get_code":{"code_sha256":"27ab42311803bf0e"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}