Research map / Label-efficient and robust learning / Recognition and long-tail
HVGCD:Rethinking Generalized Category Discovery through Hypothesis–Verification
zhang baoqiang, Kunze Huang, Luyao Tang, Xiaotong Tu
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 HVGCD | SimGCD (DINOv1) · Fine-grained average (CUB, Aircraft, Cars, Herbarium19) | All-class accuracy: 52.9 -> 56.4 | Improved |
| add HVGCD | SelEx (DINOv1) · FGVC-Aircraft | All-class accuracy: 55.2 -> 65.2 | Improved |
| add HVGCD | CMS (DINOv1) · Fine-grained average (CUB, Aircraft, Cars, Herbarium19) | All-class accuracy: 51.8 -> 53.3 | Improved |
| add HVGCD | LegoGCD (DINOv1) · Fine-grained average (CUB, Aircraft, Cars, Herbarium19) | All-class accuracy: 53.5 -> 56.3 | Improved |
| add HVGCD | SimGCD (DINOv2) · Fine-grained average (CUB, Aircraft, Cars, Herbarium19) | All-class accuracy: 66.8 -> 69.5 | Improved |
| add HVGCD | SimGCD (DINOv1) · Generic average (CIFAR-10, CIFAR-100, ImageNet-100) | All-class accuracy: 87.0 -> 88.0 | Improved |
| add HVGCD | SimGCD (DINOv1) · CIFAR-10 | All-class accuracy: 96.9 -> 96.8 | No clear effect |
| add HVGCD | SelEx (DINOv2) · CIFAR-10 | All-class accuracy: 98.3 -> 98.2 | No clear effect |
| add HVGCD | SelEx (DINOv2) · Generic average (CIFAR-10, CIFAR-100, ImageNet-100) | All-class accuracy: 92.1 -> 92.5 | No clear effect |
| add HVGCD | CMS (DINOv1) · CIFAR-100 | All-class accuracy: 78.0 -> 80.2 | Improved |
| add HVGCD | SimGCD (DINOv2) · CIFAR-100 | All-class accuracy: 86.8 -> 88.3 | Improved |
| add semantic alignment loss (gamma 0.05) (ablation) | HVGCD (SimGCD) · Average (CUB-200, Stanford-Cars, CIFAR-100) | All-class accuracy: 68.8 -> 69.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SelEx — SelEx: Self-Expertise in Fine-Grained Generalized Category Discovery (ECCV 2024) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Towards Understanding Parametric Generalized Category Discovery on Graphs (ICML 2025 · similarity 0.89)
- HIDISC: A Hyperbolic Framework for Domain Generalization with Generalized Category Discovery (NeurIPS 2025 · similarity 0.89)
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (ICLR 2025 · similarity 0.88)
- Transfer and Alignment Network for Generalized Category Discovery (AAAI 2024 · similarity 0.88)
- Hyperbolic Category Discovery (CVPR 2025 · similarity 0.88)
- SEAL: Semantic-Aware Hierarchical Learning for Generalized Category Discovery (NeurIPS 2025 · similarity 0.87)
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.
