Metraipolis › archive › dc_f207e58ffe37
After a Decade: Bringing Shadow Removal into the Real World with Agentic Training Data — arXiv:2609.38607
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.38607
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T12:43:59.387Z
- SHA-256 of the archived text
5f40c429f6d908e6cc50988aa2c258fd99784ba16d34440768d40ab374c8861c- Stable URL of this page
https://metraipolis.com/documento/dc_f207e58ffe37
How to cite
After a Decade: Bringing Shadow Removal into the Real World with Agentic Training Data — arXiv:2609.38607 — arXiv (en), https://arxiv.org/abs/2609.38607 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f207e58ffe37 (Metraipolis archive dc_f207e58ffe37, sha256 5f40c429f6d908e6cc50988aa2c258fd99784ba16d34440768d40ab374c8861c)
BibTeX
@misc{dcf207e58ffe37,
title = {After a Decade: Bringing Shadow Removal into the Real World with Agentic Training Data — arXiv:2609.38607},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.38607},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f207e58ffe37; sha256 5f40c429f6d908e6cc50988aa2c258fd99784ba16d34440768d40ab374c8861c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38607. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Shadow removal looks nearly solved on established benchmarks, yet remains brittle in the real world. Models have advanced; the paired training data they rely on have barely changed in nearly a decade. The reason is simple: obtaining a shadow-free target requires removing the occluder while keeping the scene, camera, and illumination otherwise unchanged, making diverse paired data difficult to capture. Meanwhile, large shadow detection datasets already contain diverse real-world images and masks, but no shadow-free targets. To turn this abundant but incomplete data into paired supervision, we propose an offline agentic workflow combining physics-motivated generation, failure detection, feedback-driven retry, candidate selection, and deterministic correction. Using this workflow, we construct AgenticShadow, a dataset of 17,138 image-mask-target triplets spanning general scenes, faces, and remote sensing. Our construction workflow reduces Color Distribution Difference by 50.5% over previous shadow removal work, while training existing shadow removal models on AgenticShadow reduces cross-domain LAB RMSE by 19.7-37.5%. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.38607 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T12:14:13.907Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).