AI Paper Atlas

Research map / Video / Video segmentation and tracking

Test Time Training for Long Videos via Frame Forgetting Network

Rajat modi, Xin Liang, Sebastian Noel, Yogesh Rawat

ECCV 2026 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning

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
replace sliding-window TTT-MAE with FFN (3 frames per step)Online TTT-MAE · COCO-Videosinstance segmentation AP: 37.6 -> 45.1Improved
replace sliding-window TTT-MAE with FFNOnline TTT-MAE · COCO-Videostime per frame (s): 4.1 -> 0.7Improved
replace sliding-window TTT-MAE with FFNOnline TTT-MAE · KITTI-STEP testmIoU: 54.3 -> 59.5Improved
replace sliding-window TTT-Online with FFN on hours-long videosTTT-Online (s.w.) · EpicTourssemantic segmentation mIoU: 42.8 -> 49.3Improved
replace Online TTT with FFN for action classificationOnline TTT on DINOv3 · UCF101top-1 accuracy: 94.2 -> 95.0Improved
replace Online TTT with FFN for action classificationOnline TTT on DINOv3 · SSv2top-1 accuracy: 73.4 -> 74.1Improved
add timestep conditioning of backbone (ablation)FFN · COCO-Videosinstance segmentation AP: 42.7 -> 45.1Improved
use RoPE instead of absolute time encoding (ablation)FFN · COCO-Videosinstance segmentation AP: 45.1 -> 46.8Improved
use absolute instead of relative time encoding (ablation)FFN · COCO-Videosinstance segmentation AP: 40.5 -> 45.1Improved
use MBO (merge) buffer instead of FIFO (ablation)FFN · COCO-Videosinstance segmentation AP: 43.9 -> 45.1Improved
combine FIFO and MBO buffer policies (ablation)FFN · COCO-Videosinstance segmentation AP: 45.1 -> 45.7Improved
use cosine instead of L1 for latent similarity (ablation)FFN · COCO-Videosinstance segmentation AP: 43.2 -> 45.1Improved

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