详细信息
文献类型:会议论文
英文题名:Pineapple Flower Detection Algorithm Based on Improved YOLOv5
作者:Su, Jinping[1]; Yu, Guoyan[1]
机构:[1] Guangdong Ocean University, School of Mechanical Engineering, Zhanjiang, China
会议论文集:2023 2nd International Conference on Robotics, Artificial Intelligence and Intelligent Control, RAIIC 2023
会议日期:August 11, 2023 - August 13, 2023
会议地点:Hybrid, Mianyang, China
语种:英文
外文关键词:Parameter estimation
外文摘要:Timely and accurate identification of pineapple flowers is an important step in automating the work of the flower-catalyzing robots. To accurately detect pineapple flowers in real time despite leaf occlusion or dense flowers, we propose an improved YOLOv5 algorithm. The extraction of feature information of the pineapple flower was improved by adding Convolutional Block Attention Module (CBAM) in the backbone, as well as adding a small target detection layer and using the K-means++ algorithm for anchor dimension clustering. Meanwhile, the GSConv and VoV-GSCSP-v1 modules were introduced in the neck to reduce the parameters and computation of the model. The experimental results show that the improved method improves the AP by 4.7% over the original YOLOv5 algorithm, the precision, recall, and F1 score by 2.6%, 4.8%, and 3%, respectively, and the size of the model is reduced by 0.8MB, and the number of parameters is only 89.2% of the original model. The algorithm can detect pineapple flowers more accurately and quickly in complex environments and can provide technical support for flower-catalyzing robots to execute flower catalyzing accurately and quickly in complex environments. ? 2023 IEEE.
参考文献:
正在载入数据...