Vision as Witness: Observing Actuators in an Automated Greenhouse with Vision–Language Models
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.
Authors
- Le Hoang Anh (ORCID: https://orcid.org/0000-0001-6613-056X)
- Gwang-Hyun Yu (ORCID: https://orcid.org/0000-0003-0434-5260)
- Dang Thanh Vu (ORCID: https://orcid.org/0000-0003-0743-8251)
- Jin Young Kim (ORCID: https://orcid.org/0009-0005-0551-7214)
- Jo Jung An (ORCID: https://orcid.org/0000-0002-6981-2109)
- Jaeho Song
Institutions
- Chonnam National University (KR)
Publication Details
- Journal
- Agronomy
- Published
- 2026-11-19
- DOI
- https://doi.org/10.3390/agronomy16222340
- Primary Topic
- Smart Agriculture and AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00