Research map / Video / Action recognition
Learning Goals From Failure
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Train on both intentional and unintentional action using failure-onset supervision | Our model · Oops! unintentional action classification | accuracy: 77.9 vs 64.0 | Helped |
| Represent video as contextual trajectories with a Transformer | Our model · Oops! validation SVO average | R1: 36.3 vs 33.5 | Helped |
| Apply adversarial gradient autocorrection to failed trajectories | Our model · Oops! auto-corrected trajectories, intentional SVO | Δ R5: +1.6 | Helped |
| Apply adversarial gradient autocorrection to failed trajectories | Our model · Oops! auto-corrected trajectories, unintentional SVO | Δ R5: -3.3 | Hurt |
| Linearly extrapolate the trajectory instead of adversarially correcting it | Our model · Oops! auto-corrected trajectories, intentional SVO | Δ R5: +0.6 | Helped |
| Use the Kinetics-pretrained representation rather than the proposed representation | Goal prediction representation · Oops! validation SVO object | R5: 58.2 -> 64.6 | Helped |
| Use the Kinetics-pretrained representation rather than the proposed representation | Goal prediction representation · Oops! validation all-three SVO | R5: 14.4 -> 16.5 | Helped |
Closest papers by meaning
- Oops! Predicting Unintentional Action in Video (CVPR 2020 · similarity 0.90)
- UnweaveNet: Unweaving Activity Stories (CVPR 2022 · similarity 0.87)
- Ego4D Goal-Step: Toward Hierarchical Understanding of Procedural Activities (NeurIPS 2023 · similarity 0.86)
- Grounding Video Models to Actions through Goal Conditioned Exploration (ICLR 2025 · similarity 0.86)
- Watch and Learn: Learning to Use Computers from Online Videos (CVPR 2026 · similarity 0.85)
- Can't Make an Omelette Without Breaking Some Eggs: Plausible Action Anticipation Using Large Video-Language Models (CVPR 2024 · similarity 0.84)
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.