【摘要】 目标检测是计算机视觉任务的一个重要组成部分，也是人工智能的一个重要应用，其目的是在自然图像中的大量预定义类别中定位对象实例。同时目标识别也有着广泛的应用场景，可以应用到自动驾驶技术中以及各种侦察车上。系统性介绍了主要的目标识别方法，之后主要介绍两种基于深度卷积神经网络的目标识别方法，并分别说明了它们的应用场景，同时通过具体的实验数据对比了骨干网络的优缺点。与传统的基于特征的目标检测方法相比，基于深度学习的目标检测方法可以同时学习图像的低级特征和高级特征。最后展望了目标检测方法的发展前景。更多还原

【Abstract】 This paper systematically introduces the main object recognition methods, and then mainly introduces two object recognition methods based on deep convolutional neural network, and explains their application scenarios respectively.At the same time, the advantages and disadvantages of backbone network are compared through specific experimental data. Compared with the traditional feature-based target detection method, the deep learning-based target detection method can learn both low-level features and high-level features of images. Finally, the development prospect of target detection method is prospected in this paper.更多还原