AI Paper Atlas

Research map / Video / Video-language understanding

Image-to-Video Transfer Learning based on Image-Language Foundation Models: A Comprehensive Survey

IJCV 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding

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
use LoRA fine-tuning instead of knowledge distillationCLIP video-text retrieval · MSR-VTT (video-text retrieval)R@1: 46.8 -> 50.5Improved
update CLIP parameters instead of post-network tuning on frozen featuresCLIP video-text retrieval · MSR-VTT (video-text retrieval)R@1Improved
use a stronger foundation model (Qwen2.5-VL) with simple post-network tuningtemporal video grounding · Charades-STA (TVG)mIoU: 50.9 -> 55.9Improved
use CLIP side-tuning instead of LLaVA extra-model tuningtemporal video grounding · Charades-STA (TVG)mIoU: 41.4 -> 50.9Improved
add video-level auxiliary models to MDETRspatio-temporal video grounding · VidSTG declarative (STVG)m_tIoU: 48.1 -> 52.1Improved
use LLM-based foundation model with post-network tuning instead of CLIP prompt tuningvideo question answering · NExT-QA (VideoQA)accuracy: 66.3% -> 83.73%Improved
choose adaptation strategy per benchmark with fixed CLIP backbonevideo question answering · MSRVTT-MC / How2QAaccuracy: 97.8 vs 71.4Mixed
disentangle spatial and temporal modeling (DiST instead of EVL)video action recognition · Something-Something V2 (VAR)Top-1 accuracy: 66.7 -> 73.1Improved
use adapter fine-tuning instead of side-tuningvideo action recognition · Something-Something V2 (VAR)Top-1 accuracy: 66.7 -> 70.6Improved
use knowledge distillation in open-vocabulary tracking (GLATrack)open-vocabulary MOT · TAO (OV-MOT)TETA: 37.9 vs 38.2Mixed

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