基于Mask R-CNN网络模型的无人驾驶感知

Perception of Unmanned Driving Based on Mask R-CNN

近几年深度学习技术在图像检测方面的应用取得了极大的突破,利用卷积神经网络模型可高效且准确的识别目标。一种开源网络模型——Mask R-CNN,被用于无人驾驶感知检测,取得了较好的检测效果。为了进一步提高检测精度,提出迁移学习方法重新训练网络,使得网络更适用于无人驾驶领域的感知任务。

In recent years, the application of deep learning technology in image detection has made great breakthroughs. The Convolutional Neural Network(CNN) model can be used to identify targets efficiently and accurately. An open source model--Mask R-CNN, is used for environment detection and has achieved good detection results. In order to further improve the detection accuracy, a method named transfer learning is proposed to retrain the network, making the network more suitable for the perceptive task in the unmanned driving field.