人工智能
计算机科学
特征(语言学)
目标检测
背景(考古学)
对象(语法)
趋同(经济学)
计算机视觉
特征向量
模式识别(心理学)
传感器融合
机器学习
作者
Qinxiao Liu,Zining Yan,Fang Wang,Chaoyuan Ding
标识
DOI:10.1109/cisai54367.2021.00011
摘要
With the rapid development of computer vision and artificial intelligence, the speed and real-time accuracy of detection of small agricultural pest object were significantly improved with YOLOv3, which is an object detection technology based on deep learning method. In this paper, an improved YOLOv3 detection model of fused GC Net and feature fusion is proposed. First, the proposed method can accelerate the convergence speed of the model when training, save training time by searching context feature information and lightening the network structure. Second the improved feature fusion algorithm can also enhance the accuracy of the model for small pest object by integrating the different levels of feature information in the method of vector superposition. Final the simulation shows that the mAP based on this model tested against Pest24 small object pest dataset is 65.69% compared with the original YOLOv3 model, whose average accuracy was improved by 4.27%.
科研通智能强力驱动
Strongly Powered by AbleSci AI