Research map / Detection and segmentation / Open-set and few-shot detection
Hierarchical Novelty Detection for Visual Object Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace DARTS with LOO | hierarchical novelty detection · ImageNet | novel class accuracy: 10.89 -> 15.72 | Helped |
| Replace DARTS with LOO | hierarchical novelty detection · AwA2 | novel class accuracy: 36.75 -> 50.00 | Helped |
| Replace DARTS with LOO | hierarchical novelty detection · CUB | AUC: 30.07 -> 31.92 | Helped |
| Replace Relabel with LOO (ablation) | hierarchical novelty detection · ImageNet, AwA2, and CUB | novel class accuracy: 15.29 -> 15.72; 45.71 -> 50.00; 38.23 -> 40.78 | Helped |
| Use LOO rather than relabeling, avoiding a penalty on the original label when relabeled (ablation) | hierarchical novelty detection · CUB | AUC: 28.75 -> 31.92 | Helped |
| Add top-down probability features before LOO (ablation) | LOO hierarchical novelty detection · ImageNet, AwA2, and CUB | AUC: 12.00 -> 13.98; 43.63 -> 46.77; 31.92 -> 33.16 | Helped |
| Add top-down probability features before LOO (ablation) | LOO hierarchical novelty detection · CIFAR-100 | AUC: 18.93 -> 18.54 | Hurt |
| Use relabeling instead of DARTS | hierarchical novelty detection · CUB | novel class accuracy: 40.42 -> 38.23 | Hurt |
| Add TD hierarchical embedding to attributes (ablation) | GZSL semantic embedding · AwA1 | AUC: +7.65% | Helped |
| Add TD hierarchical embedding to attributes (ablation) | GZSL semantic embedding · AwA2 | AUC: +7.97% | Helped |
| Add TD hierarchical embedding to attributes (ablation) | GZSL semantic embedding · CUB | AUC: +6.71% | Helped |
| Use TD embedding rather than Path embedding (ablation) | GZSL semantic embedding combined with other embeddings · AwA1, AwA2, and CUB | AUC | Mixed |
Closest papers by meaning
- Deep Transfer Learning for Multiple Class Novelty Detection (CVPR 2019 · similarity 0.91)
- Where's Wally Now? Deep Generative and Discriminative Embeddings for Novelty Detection (CVPR 2019 · similarity 0.89)
- Learning Deep Classifiers Consistent With Fine-Grained Novelty Detection (CVPR 2021 · similarity 0.89)
- Debiased Novel Category Discovering and Localization (AAAI 2024 · similarity 0.85)
- Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · similarity 0.85)
- Class Distribution Shifts in Zero-Shot Learning: Learning Robust Representations (NeurIPS 2024 · similarity 0.85)
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.