We’re excited to share <strong>AdaptVPR</strong>, a route-aware generative augmentation framework for improving Visual Place Recognition under challenging appearance shifts.</p>\n<p>AdaptVPR generates same-place hard positives through three scene-dependent routes:</p>\n<ul>\n<li><strong>Global Appearance Route:</strong> changes weather, illumination, and time of day.</li>\n<li><strong>Local Occlusion Route:</strong> inserts plausible dynamic occluders.</li>\n<li><strong>Dual Route:</strong> combines global and local perturbations for more challenging appearance shifts.</li>\n</ul>\n<p>A VPR-oriented verification scheme balances geometric consistency and appearance diversity, with feedback-driven prompt refinement for Local and Dual samples. Using this framework, we construct <strong>AdaptCities</strong>, a dataset of <strong>160K verified synthetic same-place hard positives</strong>.</p>\n<p>Across multiple VPR baselines and vision foundation backbones, AdaptVPR consistently improves retrieval performance.</p>\n<p>📄 <a href=\"https://arxiv.org/abs/2609.04369\" rel=\"nofollow\">Paper</a> · 💻 <a href=\"https://github.com/chenshunpeng/AdaptVPR\" rel=\"nofollow\">Code</a> · 🤗 <a href=\"https://huggingface.co/datasets/shunpeng/AdaptCities\">AdaptCities Dataset</a></p>\n","updatedAt":"2026-09-07T04:33:30.773Z","author":{"_id":"67fe8c8187b0c914e14c7994","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67fe8c8187b0c914e14c7994/e_7ES113Q6A-mBKGhSqEc.jpeg","fullname":"shunpengchen","name":"shunpeng","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7340427041053772},"editors":["shunpeng"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/67fe8c8187b0c914e14c7994/e_7ES113Q6A-mBKGhSqEc.jpeg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2609.04369","authors":[{"_id":"6a9e31e8de5ea82090db65ca","name":"Shunpeng Chen","hidden":false},{"_id":"6a9e31e8de5ea82090db65cb","name":"Jingyi Zhang","hidden":false},{"_id":"6a9e31e8de5ea82090db65cc","name":"Changwei Wang","hidden":false},{"_id":"6a9e31e8de5ea82090db65cd","name":"Shengpeng Xu","hidden":false},{"_id":"6a9e31e8de5ea82090db65ce","name":"Yukun Song","hidden":false},{"_id":"6a9e31e8de5ea82090db65cf","name":"Xingtian Pei","hidden":false},{"_id":"6a9e31e8de5ea82090db65d0","name":"Jinzhou Lin","hidden":false},{"_id":"6a9e31e8de5ea82090db65d1","name":"Li Guo","hidden":false},{"_id":"6a9e31e8de5ea82090db65d2","name":"Shibiao Xu","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/67fe8c8187b0c914e14c7994/3muXNirOJq2iwTAhutD_b.png"],"publishedAt":"2026-09-03T00:00:00.000Z","submittedOnDailyAt":"2026-09-07T00:00:00.000Z","title":"AdaptVPR: Route-Aware Hard Positive Generation for Robust Visual Place Recognition","submittedOnDailyBy":{"_id":"67fe8c8187b0c914e14c7994","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/67fe8c8187b0c914e14c7994/e_7ES113Q6A-mBKGhSqEc.jpeg","isPro":false,"fullname":"shunpengchen","user":"shunpeng","type":"user","name":"shunpeng"},"summary":"Visual Place Recognition (VPR) localizes a query image by retrieving database images of the same or nearby place, yet its robustness is often degraded by domain shifts arising from illumination, weather, seasonal changes, and dynamic occlusions. One contributing factor is the limited appearance diversity of the same place in existing training data. To address this issue, we propose AdaptVPR, a route-aware generative augmentation framework that constructs same-place hard positives for robust VPR training. AdaptVPR first uses a vision language model to parse scene attributes and estimate editing feasibility, while a rule-based scheduler determines the generation route according to editability scores and risk constraints. The generation process is decomposed into three complementary routes: the Global Appearance Route introduces global scene changes in weather, illumination, and time of day; the Local Occlusion Route inserts plausible dynamic occluders; and the Dual Route combines both types of perturbations to produce more challenging appearance shifts. Each generated candidate is evaluated using a VPR-oriented verification scheme based on geometric consistency and appearance diversity, reducing the risk of structural drift while ensuring sufficient appearance variation. Global candidates are generated once and rejected if verification fails, while Local Occlusion and Dual candidates use verification feedback for limited prompt refinement and regeneration. Using this framework, we construct AdaptCities, containing 160K verified synthetic same-place hard positives. Experiments across multiple VPR baselines and vision foundation backbones show consistent gains on standard benchmarks and substantial improvements under challenging domain shifts, with R@1 gains of up to 9.2%. The source code and data resources are publicly available at https://github.com/chenshunpeng/AdaptVPR.","upvotes":0,"discussionId":"6a9e31e8de5ea82090db65d3","githubRepo":"https://github.com/chenshunpeng/AdaptVPR","githubRepoAddedBy":"user","ai_summary":"AdaptVPR improves visual place recognition by generating verified synthetic same-place images with diverse appearance changes to train more robust models.","ai_keywords":["Visual Place Recognition","vision language model","generative augmentation","hard positives","geometric consistency","appearance diversity","domain shift"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":1},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2609/2609.04369.md","query":{}}">

Get this paper in your agent:

Don't have the latest CLI?

Cite arxiv.org/abs/2609.04369 in a model README.md to link it from this page.

Cite arxiv.org/abs/2609.04369 in a Space README.md to link it from this page.

Add this paper to a collection to link it from this page.