Research map / 3D vision / Depth and stereo
First Learn, Then Review: Human-Like Continual Learning for Cross-View Geo-Localization with Limited Field of View
Lei Cheng, Daikun Liu, Zhikun Chen, Teng Wang
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 knowledge distillation from the north-aligned panoramic teacher (ablation) | HCL-Geo (curriculum + MoE) · CVUSA, FoV 70 | R@1: 55.7 -> 67.2 | Improved |
| add knowledge distillation from the north-aligned panoramic teacher (ablation) | HCL-Geo (curriculum + MoE) · CVUSA, FoV 90 | R@1: 63.0 -> 73.8 | Improved |
| add orientation prediction module (ablation) | HCL-Geo (curriculum + KD + MoE) · CVUSA, FoV 70 | R@1: 67.2 -> 68.3 | Improved |
| add MoE review stage on top of curriculum learning (ablation) | Baseline-1 (curriculum only) · CVUSA, FoV 70 | R@1: 49.1 -> 55.7 | Improved |
| add easy-to-hard curriculum on top of MoE (ablation) | Baseline-2 (MoE only) · CVUSA, FoV 70 | R@1: 54.5 -> 55.7 | Improved |
| MoE review stage alone vs curriculum alone (ablation) | HCL-Geo · CVUSA, FoV 70 | R@1: 49.1 -> 54.5 | Improved |
| train on FoVs easy-to-hard sequentially (vanilla curriculum) without distillation (ablation) | ConvNeXt-Base · CVUSA, various FoV | R@1 | Mixed |
| increase number of training FoV tasks from four to six (ablation) | HCL-Geo · CVUSA and CVACT, FoV ranges | R@1 | No clear effect |
| use HCL-Geo instead of ConGeo trained on same FoV 90 | single-model FoV training · CVUSA, FoV 90 | R@1: 55.5 -> 71.8 | Improved |
| HCL-Geo versus ConGeo on wide FoV 180 on CVACT | single-model geo-localization · CVACT val, FoV 180 | R@1: 70.3 -> 66.4 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ConGeo — ConGeo: Robust Cross-view Geo-localization across Ground View Variations (ECCV 2024) · better on 3 of 3 shared results
- Sample4Geo — Sample4Geo: Hard Negative Sampling For Cross-View Geo-Localisation (ICCV 2023) · better on 2 of 2 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
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.88)
- Semantic Ambiguity Modeling and Propagation for Fine-Grained Visual Cross View Geo-Localization (AAAI 2025 · similarity 0.87)
- Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization (CVPR 2024 · similarity 0.87)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.86)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.85)
- Change-Robust Online Topological Memory for Long-Term Relocalization and Semantic Navigation (NeurIPS 2026 · 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.
