A prediction method for milk production of dairy cows based on fine-tuning multimodal large language models
Gangguo Song, Xueliang Chen, Dongmei Run, Mingyue Zhang, Yuhao Zhang, Wenzhe Niu, Weibin Xue · 2025
This paper proposes a method for predicting milk production in dairy cows based on fine-tuning multimodal large language models. The method utilizes the Qwen2-VL-7B-Instruct model and LoRA fine-tuning technology to process a dataset containing images and text of dairy cows. Experiments show that this method can effectively predict milk production in dairy cows. Compared with the LLaMA-7B and MiniGPT-7B models, the Qwen2-VL-7B model achieves the best F1 score of 0.93, striking a good balance between accuracy and recall rate, demonstrating its superiority in the task of predicting milk production in dairy cows and providing a scientific basis for dairy cow management.