{"about":{"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.","site":"https://codewithpapers.app","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","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/paper/continual-3d-convolutional-neural-networks","title":"Continual 3D Convolutional Neural Networks for Real-time Processing of Videos","arxiv_id":"2106.00050","date":"2021-05-31","proceeding":null,"authors":["Lukas Hedegaard","Alexandros Iosifidis"],"abstract":"We introduce Continual 3D Convolutional Neural Networks (Co3D CNNs), a new computational formulation of spatio-temporal 3D CNNs, in which videos are processed frame-by-frame rather than by clip. In online tasks demanding frame-wise predictions, Co3D CNNs dispense with the computational redundancies of regular 3D CNNs, namely the repeated convolutions over frames, which appear in overlapping clips. We show that Continual 3D CNNs can reuse preexisting 3D-CNN weights to reduce the per-prediction floating point operations (FLOPs) in proportion to the temporal receptive field while retaining similar memory requirements and accuracy. This is validated with multiple models on Kinetics-400 and Charades with remarkable results: CoX3D models attain state-of-the-art complexity/accuracy trade-offs on Kinetics-400 with 12.1-15.3x reductions of FLOPs and 2.3-3.8% improvements in accuracy compared to regular X3D models while reducing peak memory consumption by up to 48%. Moreover, we investigate the transient response of Co3D CNNs at start-up and perform extensive benchmarks of on-hardware processing characteristics for publicly available 3D CNNs.","url_abs":"https://arxiv.org/abs/2106.00050v3","url_pdf":"https://arxiv.org/pdf/2106.00050v3.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":"continual-3d-convolutional-neural-networks","repo_url":"https://github.com/LukasHedegaard/co3d","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"action-classification","task_name":"Action Classification"},{"task_slug":"video-recognition","task_name":"Video Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/action-classification-on-charades","task":"Action Classification","dataset":"Charades","model":"Co Slow_64","rank_in_archive_order":43,"of":49,"metrics":{"FLOPs (G) x views":"6.9x1","MAP":"25.2"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-charades","task":"Action Classification","dataset":"Charades","model":"Slow-8×8","rank_in_archive_order":44,"of":49,"metrics":{"FLOPs (G) x views":"54.9x1","MAP":"24.1"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-charades","task":"Action Classification","dataset":"Charades","model":"Co Slow_8","rank_in_archive_order":47,"of":49,"metrics":{"FLOPs (G) x views":"6.9x1","MAP":"21.5"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co Slow_64","rank_in_archive_order":169,"of":207,"metrics":{"Acc@1":"73.05","FLOPs (G) x views":"6.90x1","Parameters (M)":"32.45"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-L_64","rank_in_archive_order":178,"of":207,"metrics":{"Acc@1":"71.61","FLOPs (G) x views":"1.25x1","Parameters (M)":"6.15"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-M_64","rank_in_archive_order":180,"of":207,"metrics":{"Acc@1":"71.03","FLOPs (G) x views":"0.33x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"X3D-L","rank_in_archive_order":181,"of":207,"metrics":{"Acc@1":"69.29","FLOPs (G) x views":"19.17x1","Parameters (M)":"6.15"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"SlowFast-8×8-R50","rank_in_archive_order":183,"of":207,"metrics":{"Acc@1":"68.45","FLOPs (G) x views":"66.25x1","Parameters (M)":"66.25"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Slow-8x8-R50","rank_in_archive_order":186,"of":207,"metrics":{"Acc@1":"67.42","FLOPs (G) x views":"54.87x1","Parameters (M)":"32.45"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-S_64","rank_in_archive_order":187,"of":207,"metrics":{"Acc@1":"67.33","FLOPs (G) x views":"0.17x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"X3D-M","rank_in_archive_order":188,"of":207,"metrics":{"Acc@1":"67.24","FLOPs (G) x views":"4.97x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"SlowFast-4×16-R50","rank_in_archive_order":189,"of":207,"metrics":{"Acc@1":"67.06","FLOPs (G) x views":"36.46x1","Parameters (M)":"34.48"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co Slow_8","rank_in_archive_order":190,"of":207,"metrics":{"Acc@1":"65.90","FLOPs (G) x views":"6.90x1","Parameters (M)":"32.45"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"X3D-S","rank_in_archive_order":192,"of":207,"metrics":{"Acc@1":"64.71","FLOPs (G) x views":"2.06x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"I3D-R50","rank_in_archive_order":193,"of":207,"metrics":{"Acc@1":"63.98","FLOPs (G) x views":"28.61x1","Parameters (M)":"28.04"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-L_16","rank_in_archive_order":194,"of":207,"metrics":{"Acc@1":"63.03","FLOPs (G) x views":"1.25x1","Parameters (M)":"6.15"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-M_16","rank_in_archive_order":195,"of":207,"metrics":{"Acc@1":"62.80","FLOPs (G) x views":"0.33x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co X3D-S_13","rank_in_archive_order":196,"of":207,"metrics":{"Acc@1":"60.18","FLOPs (G) x views":"0.17x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co I3D_8","rank_in_archive_order":197,"of":207,"metrics":{"Acc@1":"59.58","FLOPs (G) x views":"5.68x1","Parameters (M)":"28.04"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"R(2+1)D-18_16","rank_in_archive_order":198,"of":207,"metrics":{"Acc@1":"59.52","FLOPs (G) x views":"40.71x1","Parameters (M)":"31.51"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"X3D-XS","rank_in_archive_order":199,"of":207,"metrics":{"Acc@1":"59.37","FLOPs (G) x views":"0.64x1","Parameters (M)":"3.79"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"Co I3D_64","rank_in_archive_order":200,"of":207,"metrics":{"Acc@1":"56.86","FLOPs (G) x views":"5.68x1","Parameters (M)":"28.04"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"R(2+1)D-18_8","rank_in_archive_order":201,"of":207,"metrics":{"Acc@1":"53.52","FLOPs (G) x views":"20.35x1","Parameters (M)":"31.51"},"uses_additional_data":false},{"leaderboard":"/sota/action-classification-on-kinetics-400","task":"Action Classification","dataset":"Kinetics-400","model":"RCU_8","rank_in_archive_order":202,"of":207,"metrics":{"Acc@1":"53.40","FLOPs (G) x views":"4.71x1","Parameters (M)":"12.80"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers","held":null,"reason":"arxiv_id_not_in_atlas","checked_against":{"atlas_ids":549449,"atlas_ids_sha256":"afd72399dafd9906e9062927f70655952385a3d75488558e1f0484e950ab09b7","read_at":"2026-10-05T01:46:21Z"},"by_repo":[{"repo":"LukasHedegaard/co3d","kind":"official","samples":0,"ran":0,"instrument":0,"unverified":0,"harvested_for_other_papers":1,"harvested_for":["136640364"],"harvested_for_calls":[]}],"by_repo_harvested_for_is":"harvested_for: the arXiv ids (as Syntology's graph holds them) of every paper whose samples Syntology's graph links to this repository, in arXiv-id order, complete; the page names the first 3 and says how many more. A link only: it does not say the papers are the same work, that one copied the other, or that the code implements this paper's method; harvested_for_calls gives the get_harvested_code_for_paper call for each arXiv-shaped id among them","by_repo_is":"one row per repository in the archive's code links, Syntology's graph links or the samples; samples 0 means no sample is linked to this paper, a statement about Syntology's coverage, not about the repository: with harvested_for_other_papers present, Syntology harvested the repository for that many other papers, and without it, Syntology harvested nothing from it; placed_by_identical_code counts samples whose link records no repository, placed here because identical code was harvested from this repository for another paper; ran counts executions on a synthesized input and instrument counts failures of Syntology's instrument, not of the code"},"code_links_note":null,"arxiv_metadata":null,"syntology_extracted_results":null,"record_sha256":"93427f01f4261c20bba1ed73d1e742b5d00f756f8e676e32296263c0ebe3fd39","record_changed_at":"2026-09-28","record_changed_at_basis":"first_hashed"}