Research map / Label-efficient and robust learning / Recognition and long-tail
COSTARR: Consolidated Open Set Technique with Attenuation for Robust Recognition
Ryan Rabinowitz, Steve Cruz, Walter Scheirer, Terrance E. Boult
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 |
|---|---|---|---|
| score with COSTARR (pre- plus post-attenuation features and logit) instead of PostMax | pretrained ImageNet-1K classifiers · OOSA, unknown NINCO, ResNet-50 | OOSA: 0.623 -> 0.693 | Improved |
| score with COSTARR instead of PostMax | Hiera-H · OOSA, unknown NINCO | OOSA: 0.758 -> 0.788 | Improved |
| score with COSTARR instead of PostMax | ConvNeXtV2-H · OOSA, unknown iNat | OOSA: 0.850 -> 0.856 | Improved |
| use COSTARR instead of PostMax | ResNet-50 · AUOSCR, unknown NINCO | AUOSCR: 0.624 -> 0.699 | Improved |
| add pre-attenuation features F to Hadamard-only similarity (ablation) | COSTARR · AUOSCR, ResNet-50, NINCO | AUOSCR: 0.365 -> 0.699 | Improved |
| add post-attenuation Hadamard features H to features-only similarity (ablation) | COSTARR · AUOSCR, ResNet-50, NINCO | AUOSCR: 0.681 -> 0.699 | Improved |
| scale by normalized logit and use it to select the class (ablation) | COSTARR · AUOSCR, ResNet-50, NINCO | AUOSCR: 0.690 -> 0.699 | Improved |
| scale with globally normalized logit instead of softmax (ablation) | COSTARR · AUOSCR, ResNet-50, Textures | AUOSCR: 0.725 -> 0.755 | Improved |
| use logit scaling in addition to concatenated F and H on Textures (ablation) | COSTARR · AUOSCR, ResNet-50, Textures | AUOSCR: 0.756 -> 0.755 | No clear effect |
| score with COSTARR instead of NNGuide/SCALE OOD post-processors | ResNet-50 · OOSA, unknown NINCO | OOSA: NNGuide 0.445, SCALE 0.555 -> 0.693 | Improved |
Closest papers by meaning
- All Beings Are Equal in Open Set Recognition (AAAI 2024 · similarity 0.88)
- Open-Set Recognition: A Good Closed-Set Classifier is All You Need (ICLR 2022 · similarity 0.87)
- Large-Scale Long-Tailed Recognition in an Open World (CVPR 2019 · similarity 0.87)
- Frequency Shuffling and Enhancement for Open Set Recognition (AAAI 2024 · similarity 0.85)
- Improving Open Set Recognition via Visual Prompts Distilled from Common-Sense Knowledge (AAAI 2024 · similarity 0.84)
- From Coarse to Fine-Grained Open-Set Recognition (CVPR 2024 · similarity 0.84)
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.
