dorsal/arxiv
View SchemaVIPER Strike: Defeating Visual Reasoning CAPTCHAs via Structured Vision-Language Inference
| Authors | Minfeng Qi, Dongyang He, Qin Wang, Lefeng Zhang |
|---|---|
| Categories | |
| ArXiv ID | 2601.06461vv1 |
| URL | https://arxiv.org/abs/2601.06461 |
| License | http://creativecommons.org/licenses/by/4.0/ |
Abstract
Visual Reasoning CAPTCHAs (VRCs) combine visual scenes with natural-language queries that demand compositional inference over objects, attributes, and spatial relations. They are increasingly deployed as a primary defense against automated bots. Existing solvers fall into two paradigms: vision-centric, which rely on template-specific detectors but fail on novel layouts, and reasoning-centric, which leverage LLMs but struggle with fine-grained visual perception. Both lack the generality needed to handle heterogeneous VRC deployments. We present ViPer, a unified attack framework that integrates structured multi-object visual perception with adaptive LLM-based reasoning. ViPer parses visual layouts, grounds attributes to question semantics, and infers target coordinates within a modular pipeline. Evaluated on six major VRC providers (VTT, Geetest, NetEase, Dingxiang, Shumei, Xiaodun), ViPer achieves up to 93.2% success, approaching human-level performance across multiple benchmarks. Compared to prior solvers, GraphNet (83.2%), Oedipus (65.8%), and the Holistic approach (89.5%), ViPer consistently outperforms all baselines. The framework further maintains robustness across alternative LLM backbones (GPT, Grok, DeepSeek, Kimi), sustaining accuracy above 90%. To anticipate defense, we further introduce Template-Space Randomization (TSR), a lightweight strategy that perturbs linguistic templates without altering task semantics. TSR measurably reduces solver (i.e., attacker) performance. Our proposed design suggests directions for human-solvable but machine-resistant CAPTCHAs.
{
"annotation_id": "51be08f2-039e-45aa-b349-ebd558d205e6",
"date_created": "2026-02-17T05:53:08.776000Z",
"date_modified": "2026-02-17T05:53:08.776000Z",
"file_hash": "2fe8c1fa7af3d157dc1a157a8794ac0fb0e9ce318f0884cf997c1cd0a9895edf",
"private": false,
"record": {
"abstract": "Visual Reasoning CAPTCHAs (VRCs) combine visual scenes with natural-language queries that demand compositional inference over objects, attributes, and spatial relations. They are increasingly deployed as a primary defense against automated bots. Existing solvers fall into two paradigms: vision-centric, which rely on template-specific detectors but fail on novel layouts, and reasoning-centric, which leverage LLMs but struggle with fine-grained visual perception. Both lack the generality needed to handle heterogeneous VRC deployments.\n We present ViPer, a unified attack framework that integrates structured multi-object visual perception with adaptive LLM-based reasoning. ViPer parses visual layouts, grounds attributes to question semantics, and infers target coordinates within a modular pipeline. Evaluated on six major VRC providers (VTT, Geetest, NetEase, Dingxiang, Shumei, Xiaodun), ViPer achieves up to 93.2% success, approaching human-level performance across multiple benchmarks. Compared to prior solvers, GraphNet (83.2%), Oedipus (65.8%), and the Holistic approach (89.5%), ViPer consistently outperforms all baselines. The framework further maintains robustness across alternative LLM backbones (GPT, Grok, DeepSeek, Kimi), sustaining accuracy above 90%.\n To anticipate defense, we further introduce Template-Space Randomization (TSR), a lightweight strategy that perturbs linguistic templates without altering task semantics. TSR measurably reduces solver (i.e., attacker) performance. Our proposed design suggests directions for human-solvable but machine-resistant CAPTCHAs.",
"arxiv_id": "2601.06461",
"authors": [
"Minfeng Qi",
"Dongyang He",
"Qin Wang",
"Lefeng Zhang"
],
"categories": [
"cs.CR",
"cs.CV",
"cs.ET"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"title": "VIPER Strike: Defeating Visual Reasoning CAPTCHAs via Structured Vision-Language Inference",
"url": "https://arxiv.org/abs/2601.06461",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "55409a7f-11d6-4b4c-a7c6-ecbfbaf13038",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}