AI Paper Atlas

Research map / Detection and segmentation / CNN features

Flow-Guided Feature Aggregation for Video Object Detection

Xizhou Zhu, Yujie Wang, Jifeng Dai, Lu Yuan, Yichen Wei

ICCV 2017 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 748 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 flow-guided feature aggregation with adaptive weights and end-to-end trainingR-FCN single-frame baseline · ImageNet VID val (ResNet-101)mAP: 73.4 -> 76.3Improved
add flow-guided feature aggregationR-FCN single-frame baseline · ImageNet VID val (ResNet-101, fast motion)mAP (fast): 51.4 -> 57.6Improved
add naive feature aggregation without flow or adaptive weightsR-FCN single-frame baseline · ImageNet VID val (ResNet-101)mAP: 73.4 -> 72.0Hurt
add naive feature aggregation without flowR-FCN single-frame baseline · ImageNet VID val (ResNet-101, fast motion)mAP (fast): 51.4% -> 44.6%Hurt
add adaptive weighting to naive aggregation (ablation)naive aggregation · ImageNet VID val (ResNet-101)mAP: 72.0 -> 74.3Improved
add adaptive weighting to naive aggregation (ablation)naive aggregation · ImageNet VID val (ResNet-101, fast motion)mAP (fast): 44.6% -> 52.3%Improved
add flow-guided warping to adaptive aggregation (ablation)aggregation + adaptive weights · ImageNet VID val (ResNet-101)mAP: 74.3 -> 76.3Improved
add flow-guided warping to adaptive aggregation (ablation)aggregation + adaptive weights · ImageNet VID val (ResNet-101, fast motion)mAP (fast): 52.3% -> 57.6%Improved
add end-to-end training instead of fixed single-frame networks (ablation)FGFA · ImageNet VID val (ResNet-101)mAP: 74.5 -> 76.3Improved
add Seq-NMS post-processingFGFA · ImageNet VID val (ResNet-101)mAP: 76.3 -> 78.4Improved
add MGP post-processingFGFA · ImageNet VID val (ResNet-101)mAP: 76.3 -> 75.5Hurt
add Tubelet rescoring post-processingFGFA · ImageNet VID val (ResNet-101)mAP: 76.3 -> 76.6No clear effect

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