AI Paper Atlas

Research map / Detection and segmentation / CNN features

End-to-End Human-Gaze-Target Detection With Transformers

Danyang Tu, Xiongkuo Min, Huiyu Duan, Guodong Guo, Guangtao Zhai, Wei Shen

CVPR 2022 · Detection and segmentation › CNN features · approach: RGB salient object detection · 79 citations (OpenAlex)

Check 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 HGTTR with ResNet-50 instead of VideoAttentionGazeFollowing (Real) · GazeFollowing (Real)mAP: 6.4 mAP gainImproved
use HGTTR instead of VideoAttentionVideoAttentionTarget · VideoAttentionTargetmAP: 10.3 mAP gainImproved
use HGTTR with ResNet-101 instead of DAMVideoAttentionTarget (Real) · VideoAttentionTarget (Real)L2 Dist.: 0.108 -> 0.126Hurt
set matching weights to β1=2, β2=1, β3=1, β4=2 rather than equal weights (ablation)HGTTR · VideoAttentionTargetAUC: 0.864 -> 0.893Improved
set matching weights to β1=2, β2=1, β3=1, β4=2 rather than equal weights (ablation)HGTTR · VideoAttentionTargetL2 Dist.: 0.142 -> 0.137Improved
set matching weights to β1=2, β2=1, β3=1, β4=2 rather than equal weights (ablation)HGTTR · VideoAttentionTargetmAP: 0.492 -> 0.514Improved
increase classification matching-cost weights relative to localization (ablation)HGTTR · VideoAttentionTargetAUC: 0.893 -> 0.857Hurt
increase classification matching-cost weights relative to localization (ablation)HGTTR · VideoAttentionTargetL2 Dist.: 0.137 -> 0.148Hurt
increase classification matching-cost weights relative to localization (ablation)HGTTR · VideoAttentionTargetmAP: 0.514 -> 0.487Hurt
increase the number of decoder layers (ablation)HGTTR · VideoAttentionTargetAUCImproved
increase the number of decoder layers (ablation)HGTTR · VideoAttentionTargetL2 Dist.Improved
fine-tune the head detector with distortion-augmented training data (ablation)HGTTR · VideoAttentionTarget (Blur)AUC: 0.806 -> 0.872Improved

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