Research map / Detection and segmentation / Open-set and few-shot detection
A Self Validation Network for Object-Level Human Attention Estimation
Zehua Zhang, Chen Yu, David Crandall
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 changed | Where | Measured | Effect |
|---|---|---|---|
| remove Self Validation Module in training and testing (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 37.18 | Hurt |
| remove self validation only at test time (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 41.18 | Hurt |
| test with only what->where validation (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 43.88 | Hurt |
| add SV at test to model trained without it (ablation) | Mr. Net · ATT | mAcc: 37.18 -> 39.48 | Improved |
| hard argmax vs soft argmax at test (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 44.13 | No clear effect |
| use softmax attention form for what->where validation instead of element-wise sum (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 43.30 | Hurt |
| use single optical-flow stream instead of two streams (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 37.60 | Hurt |
| use single RGB stream instead of two streams (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 42.48 | Hurt |
| add SV at test to flow-only model trained without SV (ablation) | Mr. Net flow stream · ATT | mAcc: 18.40 -> 25.10 | Improved |
| multi anchor matching instead of one-best anchor matching (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 44.27 | Hurt |
| place global class predictor after conv block 8 instead of fused block 5 (ablation) | Mr. Net · ATT | mAcc: 44.78 -> 43.69 | Hurt |
| switch from offline to online detection | Mr. Net · ATT | mAcc: 44.78 -> 39.04 | Hurt |
Closest papers by meaning
- Automatic Gaze Analysis: A Survey of Deep Learning Based Approaches (TPAMI 2023 · similarity 0.89)
- Multi-scale Object-Aware Gaze Estimation via Geometric Reasoning (ECCV 2026 · similarity 0.88)
- In the Eye of Beholder: Joint Learning of Gaze and Actions in First Person Video (ECCV 2018 · similarity 0.88)
- Connecting Gaze, Scene, and Attention: Generalized Attention Estimation via Joint Modeling of Gaze and Scene Saliency (ECCV 2018 · similarity 0.88)
- Gaze-VLM: Bridging Gaze and VLMs through Attention Regularization for Egocentric Understanding (NeurIPS 2025 · similarity 0.86)
- Learning from Observer Gaze: Zero-Shot Attention Prediction Oriented by Human-Object Interaction Recognition (CVPR 2024 · similarity 0.86)
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.
