Papers › Learn to cycle: Time-consistent feature discovery for action recognition
Learn to cycle: Time-consistent feature discovery for action recognition
Alexandros Stergiou, Ronald Poppe
Generalizing over temporal variations is a prerequisite for effective action recognition in videos. Despite significant advances in deep neural networks, it remains a challenge to focus on short-term discriminative motions in relation to the overall performance of an action. We address this challenge by allowing some flexibility in discovering relevant spatio-temporal features. We introduce Squeeze and Recursion Temporal Gates (SRTG), an approach that favors inputs with similar activations with potential temporal variations. We implement this idea with a novel CNN block that uses an LSTM to encapsulate feature dynamics, in conjunction with a temporal gate that is responsible for evaluating the consistency of the discovered dynamics and the modeled features. We show consistent improvement when using SRTG blocks, with only a minimal increase in the number of GFLOPs. On Kinetics-700, we perform on par with current state-of-the-art models, and outperform these on HACS, Moments in Time, UCF-101 and HMDB-51.
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):
get_citation_path(paper_1="2006.08247", paper_2="…")with another paper's arXiv id or titleget_concepts_for_paper(arxiv_id="2006.08247")
Code
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
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Action Classification | Kinetics-700 | SRTG r3d-101 | Top-1 Accuracy | 56.46 | #31 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r3d-101 | Top-5 Accuracy | 76.82 | #31 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r(2+1)d-50 | Top-1 Accuracy | 54.17 | #32 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r(2+1)d-50 | Top-5 Accuracy | 74.62 | #32 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r3d-50 | Top-1 Accuracy | 53.52 | #33 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r3d-50 | Top-5 Accuracy | 74.17 | #33 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r(2+1)d-34 | Top-1 Accuracy | 49.43 | #35 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r(2+1)d-34 | Top-5 Accuracy | 73.23 | #35 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r3d-34 | Top-1 Accuracy | 49.15 | #36 of 36 | Archive leaderboard | report |
| Action Classification | Kinetics-700 | SRTG r3d-34 | Top-5 Accuracy | 72.68 | #36 of 36 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-101 | Top 1 Accuracy | 33.56 | #17 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-101 | Top 5 Accuracy | 58.49 | #17 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r(2+1)d-50 | Top 1 Accuracy | 31.60 | #21 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r(2+1)d-50 | Top 5 Accuracy | 56.80 | #21 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-50 | Top 1 Accuracy | 30.72 | #22 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-50 | Top 5 Accuracy | 55.65 | #22 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r(2+1)d-34 | Top 1 Accuracy | 28.97 | #24 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r(2+1)d-34 | Top 5 Accuracy | 54.18 | #24 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-34 | Top 1 Accuracy | 28.55 | #25 of 29 | Archive leaderboard | report |
| Action Classification | MiT | SRTG r3d-34 | Top 5 Accuracy | 52.35 | #25 of 29 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-101 | Top 1 Accuracy | 84.33 | #3 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-101 | Top 5 Accuracy | 96.85 | #3 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-50 | Top 1 Accuracy | 83.77 | #4 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-50 | Top 5 Accuracy | 96.56 | #4 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-101 | Top 1 Accuracy | 81.66 | #5 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-101 | Top 5 Accuracy | 96.33 | #5 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-34 | Top 1 Accuracy | 80.39 | #6 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r(2+1)d-34 | Top 5 Accuracy | 94.27 | #6 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-50 | Top 1 Accuracy | 80.36 | #7 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-50 | Top 5 Accuracy | 95.55 | #7 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-34 | Top 1 Accuracy | 78.60 | #8 of 8 | Archive leaderboard | report |
| Action Recognition | HACS | SRTG r3d-34 | Top 5 Accuracy | 93.57 | #8 of 8 | 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
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