In recent years, the population of green iguanas in Taiwan has grown rapidly, causing significant damage to the local ecosystem and agricultural crops. Because their coloration closely resembles the natural environment, accurately detecting green iguanas remains a challenging task. We propose an enhanced You Only Look Once (YOLO)v8-based image recognition framework tailored for green iguana detection. By integrating a global attention mechanism into the Backbone of the YOLOv8 architecture and incorporating color feature weighting, the model’s feature extraction capabilities are significantly improved. These enhancements allow for a more accurate and reliable identification of green iguanas in complex natural settings. The proposed method offers a fully automated detection solution that supports agricultural and environmental experts in developing effective management strategies to control the green iguana population, thereby mitigating their ecological and agricultural impacts.