AI Paper Atlas

Research map / Video / Action recognition

UniMD: Towards Unifying Moment Retrieval and Temporal Action Detection

Yingsen Zeng, Yujie Zhong, Chengjian Feng, Lin Ma

ECCV 2024 · Video › Action recognition · approach: Fully supervised temporal action detection · 21 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
feed the query into the regression head (ablation)UniMD · Ego4D valTAD mAP: 21.86 -> 22.61Improved
use 3 fc layers instead of 1 in query transformation MLP (ablation)UniMD · Ego4D valMR R1@30: 12.65 -> 13.99Improved
pretrain on MR then finetune on TAD (ablation)UniMD · Ego4D valTAD mAP: 22.61 -> 22.69No clear effect
pretrain on TAD then finetune on MR (ablation)UniMD · Ego4D valMR R1@30: 13.99 -> 12.70Hurt
co-train TAD and MR with synchronized task sampling (ablation)UniMD · Ego4D valTAD mAP: 22.61 -> 24.12Improved
co-train with alternating task sampling instead of synchronized (ablation)UniMD · Ego4D valTAD mAP / MR R1@50: TAD mAP 22.61 -> 21.73; MR R1@50 9.34 -> 10.56Mixed
use loss weight 1-1 instead of 3-1 for co-training (ablation)UniMD co-training · Ego4D valTAD mAP: 24.12 -> 22.55Hurt
weight the TAD loss more heavily (3-1 vs 1-3) (ablation)UniMD co-training · Ego4D valTAD mAP: 19.61 -> 24.12Improved
co-train on only half of the TAD and MR training videos (ablation)UniMD · ActivityNet / ActivityNet-CaptionTAD mAP: 38.60 -> 38.62Improved
co-train on only a quarter of the training videos (ablation)UniMD · ActivityNet-CaptionMR R5@50: 77.28 -> 79.72Improved
increase co-training data volume from 0.25 to full (ablation)UniMD · ActivityNetTAD mAP: 36.78 -> 39.82Improved
add CLIP features to I3D featuresUniMD · Charades TADmAP: 24.06 -> 26.53Improved

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