Optimization of 3D Object Detection Based on Adam

This Optimization theory is a branch of mathematics that studies finding the most reasonable solution under certain conditions and is an important part of operations research. Optimization theory is now widely used in various fields of life, such as machine learning, finance, engineering, etc. Optimization models and methods are also widely used in transportation, national defense, communication and many other fields, so optimization theory has an important role in practice and can help people solve various practical problems. Adam algorithm is an optimization algorithm based on gradient descent, which combines the advantages of AdaGrad and RMSRrop algorithms with adaptive learning rate and momentum, can converge faster and avoid the problem of gradient disappearance or explosion, and plays an important role in the training of deep neural networks. In this paper, we use the Adam algorithm to optimize the selection of more reasonable parameters for the complex scenario of deep neural network and the difficulty of optimization, so that the neural network can accurately identify and localize 3D object targets.