【摘要】 随着深度学习的发展,目标检测技术得到很大的发展,在很多领域都得到了广泛应用;如行人检测、车辆检测等;动物检测同样是非常重要的一个应用领域;作出两点贡献,第一,创建一个包含6834张高清图像11个不同类别动物目标检测数据库;第二,在one-stage目标检测框架的基础上构建基础网络MiniNet,针对动物目标得出卓越的检测性能;并使该目标检测框架CPU实时,在Intel i5四核主频2.7GHz处理器上达到100ms每帧。更多还原

【Abstract】 With the development of deep learning, the object detection technology has been greatly developed in many areas and has been widely used, such as pedestrian detection, vehicle detection; Animal detection is also a very important application area. Makes the following two contributions: first, creates a database containing 6834 HD images with 11 different categories of animal. Second, builds the basic network MiniNet on the basis of the one-stage object detection framework and gets excellent performance for animal objects. And makes the object detection framework CPU real time, reaches 100 ms per frame in Intel i5 quad-core frequency 2.7 GHz processor.更多还原