Greenhouse anomalygpt: real-time monitoring with small language-vision models
Greenhouse AnomalyGPT leverages small language-vision models for real-time anomaly detection and description in greenhouses, providing an interpretable and user-friendly monitoring solution.
Overview
Greenhouse AnomalyGPT is an innovative solution designed to enhance real-time monitoring in greenhouse environments. By utilizing small language-vision models, this technology detects anomalies and provides understandable descriptions, significantly improving the interpretability of monitoring systems. This approach aims to streamline the anomaly detection process, making it more efficient and accessible for greenhouse operations.
Technical specifications
The solution adapts the AnomalyGPT model to a smaller scale using knowledge distillation, allowing it to run effectively on edge devices. This smaller language-vision model maintains the zero-shot capabilities of larger models while ensuring real-time performance. The system collects images from the greenhouse, processes them to detect issues, and communicates findings through a user-friendly interface. Key features include:
- Real-time anomaly detection and description
- Knowledge distillation from larger models for enhanced efficiency
- Zero-shot learning capabilities for immediate deployment
- User-friendly interface for easy interaction and feedback
Technology readiness level
The project currently stands at Technology Readiness Level 3, with the smaller language-vision model undergoing validation through quantitative performance metrics. Following successful validation, the model will be integrated into a comprehensive pipeline, tested in various greenhouse conditions, and refined using feedback mechanisms over the course of 1-2 years.
About Texas A&M University, College Station
Texas A&M University in College Station is a comprehensive public research university and the flagship of The Texas A&M University System, combining broad academic strengths with a strong applied‑research culture. Industry collaborates on the Texas A&M‑RELLIS campus—an integrated education, research and testing environment that supports large‑scale experimentation and proving grounds—and through the Texas A&M Transportation Institute’s facilities in Bryan‑College Station. A statewide extension network connects university expertise to companies and communities across all Texas counties, enabling rapid piloting and deployment. Research is supported by competitive federal funding from agencies such as NSF, NIH, DOE, USDA and DoD, alongside state and industry sponsorship. Texas A&M Innovation provides IP management, licensing and commercialization pathways across the system.
Inquire about this partnering listing
Industry applications
delivered to your inbox.