{"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/can-spatiotemporal-3d-cnns-retrace-the","title":"Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet?","arxiv_id":"1711.09577","date":"2017-11-27","proceeding":"CVPR 2018 6","authors":["Kensho Hara","Hirokatsu Kataoka","Yutaka Satoh"],"abstract":"The purpose of this study is to determine whether current video datasets have\nsufficient data for training very deep convolutional neural networks (CNNs)\nwith spatio-temporal three-dimensional (3D) kernels. Recently, the performance\nlevels of 3D CNNs in the field of action recognition have improved\nsignificantly. However, to date, conventional research has only explored\nrelatively shallow 3D architectures. We examine the architectures of various 3D\nCNNs from relatively shallow to very deep ones on current video datasets. Based\non the results of those experiments, the following conclusions could be\nobtained: (i) ResNet-18 training resulted in significant overfitting for\nUCF-101, HMDB-51, and ActivityNet but not for Kinetics. (ii) The Kinetics\ndataset has sufficient data for training of deep 3D CNNs, and enables training\nof up to 152 ResNets layers, interestingly similar to 2D ResNets on ImageNet.\nResNeXt-101 achieved 78.4% average accuracy on the Kinetics test set. (iii)\nKinetics pretrained simple 3D architectures outperforms complex 2D\narchitectures, and the pretrained ResNeXt-101 achieved 94.5% and 70.2% on\nUCF-101 and HMDB-51, respectively. The use of 2D CNNs trained on ImageNet has\nproduced significant progress in various tasks in image. We believe that using\ndeep 3D CNNs together with Kinetics will retrace the successful history of 2D\nCNNs and ImageNet, and stimulate advances in computer vision for videos. The\ncodes and pretrained models used in this study are publicly available.\nhttps://github.com/kenshohara/3D-ResNets-PyTorch","url_abs":"http://arxiv.org/abs/1711.09577v2","url_pdf":"http://arxiv.org/pdf/1711.09577v2.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":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/kenshohara/3D-ResNets-PyTorch","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/LiliMeng/3D-ResNets-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/LiliMeng/3D-ResNets-Pytorch-ImageNet-Moments","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/ShiYaya/video-classification-3d-cnn-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/Xiehuaiqi/3D-ResNets-PyTorch-master","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/ankitshah009/unsupervised_video_retrieval_from_activity","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/arundhatikurup/3DResnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/buriedms/3D-ResNets-Paddle","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"paddle","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/carVaba/video-classification-3d-cnn-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/concordiaca/3D-ResNets-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/danielmatte/3D-ResNets-PyTorch-","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/dongzhuoyao/3D-ResNets-PyTorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/generation21/generation6011","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/hjjpku/adaptive_sampler","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/kaiqiangh/extracting-video-features-ResNeXt","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/kenshohara/3D-ResNets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/kenshohara/video-classification-3d-cnn-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/liyunqianggyn/Deep-Unsupervised-Image-Hashing","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/nCheck/video-class-3d-resnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/okankop/Efficient-3DCNNs","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/qwarts/video_classification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/sebastiantiesmeyer/deeplabchop3d","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/skamdar/gesture_recognition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/tianhai123/3D-ResNets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/yangbang18/video-classification-3d-cnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"can-spatiotemporal-3d-cnns-retrace-the","repo_url":"https://github.com/justld/3D-ResNets_paddle","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"paddle","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"action-recognition-in-videos","task_name":"Action Recognition"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"grouped-convolution","method_name":"Grouped Convolution"},{"method_slug":"kaiming-initialization","method_name":"Kaiming Initialization"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"resnext","method_name":"ResNeXt"},{"method_slug":"resnext-block","method_name":"ResNeXt Block"},{"method_slug":"residual-connection","method_name":"Residual Connection"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/action-recognition-in-videos-on-ucf101","task":"Action Recognition","dataset":"UCF101","model":"ResNeXt-101 (64f)","rank_in_archive_order":52,"of":91,"metrics":{"3-fold Accuracy":"94.5"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1711.09577","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1711.09577"}},"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/tianhai123/3D-ResNets","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/arundhatikurup/3DResnet","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/concordiaca/3D-ResNets-PyTorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Xiehuaiqi/3D-ResNets-PyTorch-master","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/qwarts/video_classification","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/dongzhuoyao/3D-ResNets-PyTorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/generation21/generation6011","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nCheck/video-class-3d-resnet","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/carVaba/video-classification-3d-cnn-pytorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kaiqiangh/extracting-video-features-ResNeXt","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/skamdar/gesture_recognition","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ShiYaya/video-classification-3d-cnn-pytorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/okankop/Efficient-3DCNNs","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/justld/3D-ResNets_paddle","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/danielmatte/3D-ResNets-PyTorch-","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/LiliMeng/3D-ResNets-Pytorch-ImageNet-Moments","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kenshohara/3D-ResNets","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kenshohara/3D-ResNets-PyTorch","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sebastiantiesmeyer/deeplabchop3d","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/buriedms/3D-ResNets-Paddle","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/liyunqianggyn/Deep-Unsupervised-Image-Hashing","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/kenshohara/video-classification-3d-cnn-pytorch","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/yangbang18/video-classification-3d-cnn","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ankitshah009/unsupervised_video_retrieval_from_activity","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/LiliMeng/3D-ResNets-PyTorch","reach":{"status":"unanswered"}}],"summary":{"ran_draft_wrong":7,"unverified":1},"by_repo_kind":{"listed":{"samples":5,"ran":5,"repositories":3}},"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":"213ccc8fb87ee8d9","entry":"conv1x1x1","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":"213ccc8fb87ee8d9"}},{"code_sha256_prefix":"c3b215555357b9d1","entry":"conv3x3x3","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":"c3b215555357b9d1"}},{"code_sha256_prefix":"43c0211ff8dbca5f","entry":"downsample_basic_block","repo":"tianhai123/3D-ResNets","repo_kind":"listed","path":"models/resnet.py","file_url":"https://github.com/tianhai123/3D-ResNets/blob/HEAD/models/resnet.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"43c0211ff8dbca5f"}},{"code_sha256_prefix":"93801ecf3026ab06","entry":"get_fine_tuning_parameters","repo":"tianhai123/3D-ResNets","repo_kind":"listed","path":"models/resnet.py","file_url":"https://github.com/tianhai123/3D-ResNets/blob/HEAD/models/resnet.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"93801ecf3026ab06"}},{"code_sha256_prefix":"8d57521c7b52b777","entry":"get_fine_tuning_parameters","repo":"arundhatikurup/3DResnet","repo_kind":"listed","path":"models/resnet.py","file_url":"https://github.com/arundhatikurup/3DResnet/blob/HEAD/models/resnet.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8d57521c7b52b777"}},{"code_sha256_prefix":"8484994a54d4c487","entry":"get_fine_tuning_parameters","repo":"okankop/Efficient-3DCNNs","repo_kind":"listed","path":"models/resnet.py","file_url":"https://github.com/okankop/Efficient-3DCNNs/blob/HEAD/models/resnet.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"8484994a54d4c487"}},{"code_sha256_prefix":"90b941a8c18bcd57","entry":"load_labels","repo":"okankop/Efficient-3DCNNs","repo_kind":"listed","path":"utils/kinetics_json.py","file_url":"https://github.com/okankop/Efficient-3DCNNs/blob/HEAD/utils/kinetics_json.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"90b941a8c18bcd57"}},{"code_sha256_prefix":"392a1ed5b8624559","entry":"generate_model","repo":null,"repo_kind":null,"path":null,"file_url":null,"link_basis":"identical_code_first_harvested_elsewhere","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":null,"inline_ok":false,"mcp_get_code":{"code_sha256":"392a1ed5b8624559"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}