Vision as Witness: Observing Actuators in an Automated Greenhouse with Vision–Language Models
Agronomy · 19.11.2026 · 16 (22) · 2340
DOI: 10.3390/agronomy16222340 ↗Kurz gefasst
Fachartikel in der Zeitschrift Agronomy, online erschienen am 19.11.2026. Verfasst von Le Hoang Anh und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Research Program for Agriculture Science and Technology Development, Ministry of Agriculture, Food and Rural Affairs, Ministry of Science and ICT (MSIT, KoreCitya) and Gwangju Metropolitan.
Abstract (Originalsprache)
Greenhouse controllers record commands but cannot confirm whether actuators execute them. We propose Vision as Witness, a framework that uses vision-language models (VLMs) to infer actuator states from greenhouse video and treat camera observations as visual evidence of actuator behavior. To support this task, we introduce the Naju Actuator Vision-Log Dataset (NAVL), comprising 242 days of controller logs, 149.8 h of video from two cameras, and 1000 manually annotated segments covering irrigation, circulation fans, and a shading curtain. Analysis of the 75 paired days shows that fixed-schedule recording captured only 27.6% of 1557 commands. Zero-shot VLMs perform near chance level (44.5–60.3% balanced accuracy), whereas a spectral motion detector reads the fan at 73.0% and reference photographs in the prompt raise irrigation from 50.0% to 63.0%. With class-balanced LoRA fine-tuning, curtain, fan and irrigation states are read at 75.8–88.1%, 71.4–90.5% and 79.4–95.5% on held-out dates. Importantly, video surpasses a single frame by 52.2, 29.0 and 54.2 points on curtain motion, fan and irrigation in Qwen2.5-VL-7B, at 13 times the prompt tokens. These results suggest that greenhouse cameras can serve as independent witnesses for some actuator states, although they require adaptation for robust verification.
Förderung
- Research Program for Agriculture Science and Technology Development
- Ministry of Agriculture, Food and Rural Affairs
- Ministry of Science and ICT (MSIT, KoreCitya) and Gwangju Metropolitan
Zitieren
APA
Anh, L. H., Vu, D. T., Song, J., Kim, J. Y., An, J. J., & Yu, G. H. (2026). Vision as Witness: Observing Actuators in an Automated Greenhouse with Vision–Language Models. Agronomy, 16(22), 2340. https://doi.org/10.3390/agronomy16222340
Vancouver
Anh LH, Vu DT, Song J, Kim JY, An JJ, Yu GH. Vision as Witness: Observing Actuators in an Automated Greenhouse with Vision–Language Models. Agronomy. 2026;16(22):2340. doi:10.3390/agronomy16222340
BibTeX
@article{Anh2026,
author = {Anh, Le Hoang and Vu, Dang Thanh and Song, Jaeho and Kim, Jin Young and An, Jo Jung and Yu, Gwang Hyun},
title = {Vision as Witness: Observing Actuators in an Automated Greenhouse with Vision–Language Models},
journal = {Agronomy},
year = {2026},
volume = {16},
number = {22},
pages = {2340},
doi = {10.3390/agronomy16222340}
}
EUVIMED