AI Paper Atlas

Research map / Video / Video segmentation and tracking

Rethinking Image-to-Video Adaptation: An Object-centric Perspective

Rui Qian, Shuangrui Ding, Dahua Lin

ECCV 2024 · Video › Video segmentation and tracking · approach: Mask propagation for unsupervised video segmentation · 5 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
add object distillation loss L_obj (ablation)Object-centric image-to-video adaptation · Kinetics-400Top-1 accuracy: 70.4 -> 80.3Improved
add temporal reasoning loss L_temp (ablation)Object-centric image-to-video adaptation · Something-Something-v2Top-1 accuracy: 44.8 -> 52.1Improved
add both L_obj and L_temp (ablation)Object-centric image-to-video adaptation · Something-Something-v2Top-1 accuracy: 44.8 -> 71.6Improved
add L_temp (ablation)Object-centric image-to-video adaptation · DAVIS-2017-UVOSJ&F: 27.3 -> 30.5Improved
compress frames into 8 object tokens instead of keeping all original HW features (ablation)Temporal modeling input · Something-Something-v2Top-1 accuracy: 64.3 -> 71.2Improved
increase object tokens from 8 to 16 (ablation)Slot attention · Kinetics-400 / SSv2Top-1 accuracy: 85.8 -> 85.9 (K-400); 71.2 -> 71.2 (SSv2)No clear effect
increase object tokens from 1 to 8 (ablation)Slot attention · Something-Something-v2Top-1 accuracy: 44.1 -> 71.2Improved
add temporal attention module on top of object-time interaction (ablation)Object-time interaction · Something-Something-v2Top-1 accuracy: 71.6 -> 66.8Hurt
add object state change modeling (Phi_ti) to object interaction module (ablation)Object-time interaction · Something-Something-v2Top-1 accuracy: 50.0 -> 71.6Improved
vary sampling interval delta from 2 to larger values (or all intervals) (ablation)State change modeling · Something-Something-v2Top-1 accuracy: 71.2 (delta=2) vs 71.2 (All)No clear effect
apply CRF refinement to zero-shot segmentation masksOurs K-400 · DAVIS-2017-UVOSJ&F: 44.7 -> 49.3Improved
train object tokens on K-400 rather than SSv2 (ablation)Zero-shot video object segmentation · DAVIS-2017-UVOSJ&F: 41.0 -> 44.7Improved

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