AI Paper Atlas

Research map / Video / Action recognition

P3IV: Probabilistic Procedure Planning From Instructional Videos With Weak Supervision

He Zhao, Isma Hadji, Nikita Dvornik, Konstantinos G. Derpanis, Richard P. Wildes, Allan D. Jepson

CVPR 2022 · Video › Action recognition · approach: Action anticipation and latent actions · 25 citations (OpenAlex)

Check 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
Add language-based contrastive supervision to action supervision (ablation)P3IV · CrossTask T=3SR: 16.90 -> 22.12Improved
Add adversarial loss to action and language supervision (ablation)P3IV · CrossTask T=3SR: 22.12 -> 23.34Improved
Use Viterbi post-processing (ablation)P3IV · NIVSR: 20.18 -> 24.68Improved
Use a learned memory of size 128 rather than no memory (ablation)P3IV · CrossTask T=3SR: 7.49 -> 23.34Improved
Increase memory size from 128 to 256 (ablation)P3IV · CrossTask T=3SR: 23.34 -> 20.81Hurt
Use a probabilistic adversarial model rather than a deterministic model (ablation)P3IV · CrossTask T=3KL-Div: 2.31 -> 2.11Improved
Add diversity regularization to probabilistic training (ablation)P3IV · CrossTask T=3NLL: 5.07 -> 4.89Improved
Use HowTo100M-pretrained joint video-language features instead of CrossTask-trained features (ablation)P3IV · CrossTask T=3SR: 10.82 -> 23.34Improved
Replace weak language supervision with strong visual supervision (ablation)P3IV · CrossTask T=3SR: 23.34 -> 24.41Mixed
Increase inference samples from 150 to 2500 (ablation)P3IV probabilistic · CrossTask T=3NLL: 5.03 ± 0.21 -> 4.89 ± 0.04Improved
Use a probabilistic rather than deterministic model (ablation)P3IV · COINNLL: 6.37 -> 6.37No clear effect

Later papers that compared with this method

3 later accepted papers report P3IV as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.

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