AI Paper Atlas

Research map / Video / Action recognition

Skip-Plan: Procedure Planning in Instructional Videos via Condensed Action Space Learning

Zhiheng Li, Wenjia Geng, Muheng Li, Lei Chen, Yansong Tang, Jiwen Lu, Jie Zhou

ICCV 2023 · Video › Action recognition · approach: Action anticipation and latent actions · 7 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
skip intermediate language state supervision (ablation)Skip-Plan · CrossTask (T=3)Success Rate: 26.28 -> 28.85Improved
use shared MLP on time dimension of input frames instead of average pooling (ablation)Skip-Plan · CrossTask (T=3)Success Rate: 24.40 -> 28.85Improved
use MLP instead of 1D conv on input frame time (ablation)Skip-Plan · CrossTask (T=3)Success Rate: 26.56 -> 28.85Improved
decompose long chains into sub-chains (ablation)Skip-Plan · CrossTask (T=4)Success Rate: 12.72 -> 15.56Improved
decompose long chains into sub-chains (ablation)Skip-Plan · CrossTask (T=5)Success Rate: 4.97 -> 8.55Improved
decompose long chains into sub-chains (ablation)Skip-Plan · CrossTask (T=6)Success Rate: 2.62 -> 5.12Improved
use focal loss gamma 1.5 instead of cross-entropy (gamma 0) (ablation)Skip-Plan · CrossTask (T=4)Success Rate: 13.84 -> 15.56Improved
train sub-chain alone (short loss) vs inside the full chain loss (ablation)Skip-Plan sub-chain {a1,a3,a5} · CrossTask (T=5)Success Rate: 13.73 -> 16.29Improved
use adjacent unreliable sub-chain {a1,a2,a3} instead of sub-chains anchored on first and last actions (ablation)Skip-Plan decoupling · CrossTask (T=4)Success Rate: 16.65 -> 15.91Hurt
Skip-Plan vs P3IV on CrossTask T=4 success rateSkip-Plan · CrossTask (T=4)Success Rate: 16.41 (Ext-GAIL) vs 15.56Mixed

Later papers that compared with this method

1 later accepted paper reports Skip-Plan as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy