Reproduce “PlantVLM-Edge: Quantized Vision-Language Models for On-Device Crop Health Diagnosis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Gaurab Wagle, Bishal Adhikari, Jiajia Li, Xiaolei Huang, Dong Chen
PlantVLM-Edge adapts vision-language models for offline agricultural question answering, so growers can ask follow-up questions without sending crop images to the cloud. The authors fine-tune language backbones with LoRA, then quantize their linear layers and serve them locally. On Jetson AGX Orin, Qwen3-VL-2B W4A16-AWQ retained 98.6% of full-precision ROUGE-L, a text-overlap score, not diagnostic accuracy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.