Obstacle Detection with Deep Convolutional Neural Network
Hong Yu, Ruxia Hong, Xiaolei Huang, Zhengyou Wang · 2013
The difficulty of obstacle detection is how to locate and separate the obstacle from the complex background. Traditional computer vision algorithms can not handle this problem very well due to the handcrafted designed features are vulnerable in complex background. In this article, we use deep convolutional neural network (CNN) to detect obstacle in complex scene. The deep architecture of the CNN guarantees the features learned by the network are rich and effective for detecting the obstacle. The results show that the model achieved a good performance.