AI Paper Atlas

Research map / Detection and segmentation / CNN features

Efficient One-Stage Video Object Detection by Exploiting Temporal Consistency

Guanxiong Sun, Yang Hua, Guosheng Hu, Neil Robertson

ECCV 2022 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 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
add the location prior network (ablation)FCOS · ImageNet VID valAP: 49.8 -> 54.1Improved
add the location prior network (ablation)CenterNet · ImageNet VID valAP: 49.3 -> 53.4Improved
add the location prior network (ablation)YOLOX-M · ImageNet VID valAP: 49.4 -> 53.3Improved
add the location prior network (ablation)FCOS · ImageNet VID valFPS: 25.1 -> 20.4Hurt
add the size prior network (ablation)FCOS+LPN · ImageNet VID valFPS: 20.4 -> 26.9Improved
add the size prior network (ablation)FCOS+LPN · ImageNet VID valAP: 54.1 -> 53.8Mixed
add the size prior network (ablation)YOLOX-M+LPN · ImageNet VID valFPS: 35.8 -> 50.5Improved
add the size prior network (ablation)YOLOX-M+LPN · ImageNet VID valAP: 53.3 -> 52.7Mixed
raise box adjustment ratio r from 0.5 to 0.8 (ablation)FCOS+LPN · ImageNet VID valAP: 53.6 -> 54.1Improved
raise box adjustment ratio r from 0.8 to 1.5 (ablation)FCOS+LPN · ImageNet VID valFPS: 20.4 -> 14.7Hurt
raise frame interval T from 7 to 28 (ablation)FCOS+LPN+SPN · ImageNet VID valAP: 53.8 -> 48.5Hurt
use attention on all pixels instead of 300 proposals (naive one-stage adaptation)SELSA-style attention · ImageNet VID, 2 reference framesFPS: 18.5 -> 4.6Hurt

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