摘要
行人检测是智能交通视频监控领域的一项基础技术。本文采用基于视网膜理论的图像增强策略对训练样本进行预处理,以减小光照变化的影响。首先,我们使用图像增强方法来增强图像的对比度。然后,我们将原始样本放入带有YOLOv3的darknet帧中训练检测模型1,将增强后的样本放入YOLOv3中训练检测模型2。最后,我们用200张行人照片对这两种模型进行了测试。实验结果表明,经视网膜x图像增强训练的模型具有94%的准确率。
Pedestrian detection is a basic technology in the field of intelligent traffic video surveillance. In this paper, image enhancement strategy based on retina theory is used to preprocess training samples to reduce the influence of light changes. Firstly, we use image enhancement to enhance the contrast of the image. Then, we put the original sample into the Darknet frame with YOLOv3 to train the detection model 1, and the enhanced sample into YOLOv3 to train the detection model 2. Finally, we tested the two models with 200 pedestrian photos. The experimental results show that the model trained by retinal X-ray image enhancement has 94% accuracy.
作者
郭鸣宇
刘实
Guo Mingyu;Liu Shi(Shenyang City College,Shenyang Liaoning,110112)
出处
《电子测试》
2019年第15期65-66,86,共3页
Electronic Test
