Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2601.03011 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915710918197248 |
|---|---|
| author | Wei, Yihan Yuan, Shenghai Deng, Tianchen Lou, Boyang Hu, Enwen |
| author_facet | Wei, Yihan Yuan, Shenghai Deng, Tianchen Lou, Boyang Hu, Enwen |
| contents | Corner cases are rare or extreme scenarios that drive real-world failures, but they are difficult to curate at scale: web data are noisy, labels are brittle, and edge deployments preclude large retraining. We present ReCCur (Recursive Corner-Case Curation), a low-compute framework that converts noisy web imagery into auditable fine-grained labels via a multi-agent recursive pipeline. First, large-scale data acquisition and filtering expands a domain vocabulary with a vision-language model (VLM), crawls the web, and enforces tri-modal (image, description, keyword) consistency with light human spot checks to yield refined candidates. Next, mixture-of-experts knowledge distillation uses complementary encoders (e.g., CLIP, DINOv2, BEiT) for kNN voting with dual-confidence activation and uncertainty sampling, converging to a high-precision set. Finally, region-evidence VLM adversarial labeling pairs a proposer (multi-granularity regions and semantic cues) with a validator (global and local chained consistency) to produce explainable labels and close the loop. On realistic corner-case scenarios (e.g., flooded-car inspection), ReCCur runs on consumer-grade GPUs, steadily improves purity and separability, and requires minimal human supervision, providing a practical substrate for downstream training and evaluation under resource constraints. Code and dataset will be released. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_03011 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ReCCur: A Recursive Corner-Case Curation Framework for Robust Vision-Language Understanding in Open and Edge Scenarios Wei, Yihan Yuan, Shenghai Deng, Tianchen Lou, Boyang Hu, Enwen Computer Vision and Pattern Recognition Multiagent Systems Corner cases are rare or extreme scenarios that drive real-world failures, but they are difficult to curate at scale: web data are noisy, labels are brittle, and edge deployments preclude large retraining. We present ReCCur (Recursive Corner-Case Curation), a low-compute framework that converts noisy web imagery into auditable fine-grained labels via a multi-agent recursive pipeline. First, large-scale data acquisition and filtering expands a domain vocabulary with a vision-language model (VLM), crawls the web, and enforces tri-modal (image, description, keyword) consistency with light human spot checks to yield refined candidates. Next, mixture-of-experts knowledge distillation uses complementary encoders (e.g., CLIP, DINOv2, BEiT) for kNN voting with dual-confidence activation and uncertainty sampling, converging to a high-precision set. Finally, region-evidence VLM adversarial labeling pairs a proposer (multi-granularity regions and semantic cues) with a validator (global and local chained consistency) to produce explainable labels and close the loop. On realistic corner-case scenarios (e.g., flooded-car inspection), ReCCur runs on consumer-grade GPUs, steadily improves purity and separability, and requires minimal human supervision, providing a practical substrate for downstream training and evaluation under resource constraints. Code and dataset will be released. |
| title | ReCCur: A Recursive Corner-Case Curation Framework for Robust Vision-Language Understanding in Open and Edge Scenarios |
| topic | Computer Vision and Pattern Recognition Multiagent Systems |
| url | https://arxiv.org/abs/2601.03011 |