Research map / Graph learning / Knowledge and temporal graphs
Hallucinating Visual Instances in Total Absentia
Jiayan Qiu, Yiding Yang, Xinchao Wang, Dacheng Tao
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 |
|---|---|---|---|
| add global autoencoder feature (ablation) | GCN module · COCO | accuracy (%): 86.55 -> 87.93 | Improved |
| add global autoencoder feature (ablation) | GNN baseline · COCO | accuracy (%): 75.67 -> 79.23 | Improved |
| add global autoencoder feature (ablation) | GCN module · COCO | balanced accuracy (%): 84.19 -> 85.62 | Improved |
| add spatial coordinates to node features (ablation) | GCN module · COCO | accuracy (%): 84.86 -> 87.93 | Improved |
| add spatial coordinates to node features (ablation) | GNN baseline · COCO | accuracy (%): 77.64 -> 79.23 | Improved |
| use cross-entropy instead of KL divergence loss (ablation) | GCN module · COCO | accuracy (%): 85.46 -> 87.93 | Improved |
| use cross-entropy instead of KL divergence loss (ablation) | GNN baseline · COCO | accuracy (%): 73.01 -> 79.23 | Improved |
| use attention-based GCN instead of Kipf GNN (ablation) | class prediction · COCO | accuracy (%): 79.23 -> 87.93 | Improved |
| train OGAN jointly with GRAN end-to-end (ablation) | object generator · GAN pre-training | PSNR: 26.23 -> 27.14 | Improved |
| train refinement network jointly end-to-end (ablation) | refinement network · GAN pre-training | PSNR: 29.06 -> 29.17 | Improved |
| train OGAN jointly with GRAN end-to-end (ablation) | object generator · GAN pre-training | SSIM: 0.9002 -> 0.9078 | Improved |
| train GCN and GAN modules end-to-end rather than separately (ablation) | full framework · COCO | visual plausibility | Improved |
Closest papers by meaning
- Same Attention, Different Truths: Put Logit-Lens over Visual Attention to Detect and Mitigate LVLM Object Hallucination (CVPR 2026 · similarity 0.87)
- Analyzing and Mitigating Object Hallucination: A Training Bias Perspective (AAAI 2026 · similarity 0.86)
- Mitigating Object Hallucination in Large Vision-Language Models via False Discovery Controlled Visual Data Splitting (NeurIPS 2026 · similarity 0.86)
- HulluEdit: Single-Pass Evidence-Consistent Subspace Editing for Mitigating Hallucinations in Large Vision-Language Models (CVPR 2026 · similarity 0.86)
- Finding the Correct Visual Evidence Without Forgetting: Mitigating Hallucination in LVLMs via Inter-Layer Visual Attention Discrepancy (ICML 2026 · similarity 0.86)
- VASparse: Towards Efficient Visual Hallucination Mitigation via Visual-Aware Token Sparsification (CVPR 2025 · 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.
