APPLICATION OF FAST SEGMENT ANYTHING MODEL (FASTSAM) FOR AUTONOMOUS ROBOT IDENTIFYING PLANT DISEASE

Tóm tắt

Large language models like Fast Segment Anything Model (FastSAM) have shown promising capabilities in few-shot learning across diverse domains. In this paper, we explore the application of FastSAM for plant disease identification by autonomous robots utilizing simultaneous localization and mapping (SLAM). We propose fine-tuning FastSAM on a dataset of plant images labelled with different disease types. The fine-tuned model is then deployed on an autonomous robot equipped with cameras and SLAM capabilities to identify plant diseases in real-world agricultural settings. Our results demonstrate that FastSAM can accurately recognize multiple plant diseases after being fine-tuned with only a few examples per class. The approach allows reliable plant disease identification without extensive training in data collection. This research highlights the potential of large language models like FastSAM for practical autonomous robot applications like precision agriculture when combined with technologies like SLAM