AI Paper Atlas

Research map / Video / Video segmentation and tracking

Accurate and Fast Compressed Video Captioning

Yaojie Shen, Xin Gu, Kai Xu, Heng Fan, Longyin Wen, Libo Zhang

ICCV 2023 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 67 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 motion vector input with action encoder to I-frame (ablation)CoCap · MSRVTTCIDEr: 54.1 -> 55.3Improved
add residual input on top of I-frame and motion vector (ablation)CoCap · MSRVTTCIDEr: 55.3 -> 56.2Improved
add the action encoder (ablation)CoCap with I+MV+Res · MSRVTTCIDEr: 54.3 -> 56.2Improved
use only motion vector or residual without I-frame (ablation)CoCap · MSRVTTCIDEr: 54.1 -> 19.4 (MV), 13.0 (Res)Hurt
use smaller GOP size (KeyInt 60 instead of 250) (ablation)CoCap · MSRVTTCIDEr: 49.5 -> 52.4Improved
increase sampled GOPs from 2 to 8 (ablation)CoCap · MSRVTTCIDEr: 52.4 -> 56.2Improved
increase sampled GOPs from 8 to 10 (ablation)CoCap · MSRVTTCIDEr: 56.2 -> 55.5Hurt
use deeper I-frame encoder (24 layers instead of 12) (ablation)CoCap · MSRVTTCIDEr: 56.2 -> 57.2Improved
add more layers to motion, residual, action encoders or decoder (ablation)CoCap · MSRVTTCIDEr: 56.2 -> 55.2No clear effect
use compressed-domain input instead of decoded RGB frames for speedCoCap vs SwinBERT · MSRVTTtotal inference time (ms): 339 -> 178Improved
use CLIP-pretrained backbone instead of VidSwin pretrained on Kinetics-600CoCap vs SwinBERT · VATEXCIDEr: 73 -> 52.7Hurt

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