Search Results

Image segmentation is critical in autonomous driving field. It can reveal essential clues such as objects’ shape or boundary information. The information, moreover, can be leveraged as input information of other tasks: vehicle detection, for example, or vehicle trajectory prediction. SegNet, one deep learning based segmentation model proposed by Cambridge, has been a public baseline for scene perception tasks. It, however, suffers an accuracy deficiency in objects marginal area. Segmentation of this area is very challenging with current models. To alleviate the problem, in this paper, we propose one edge enhanced deep learning based model. Specifically, we first introduced one simple, yet effective Artificial Interfering Mechanism (AIM) which feeds segmentation model manual extracted key features. We argue this mechanism possesses the ability to enhance essential features extraction and hence, ameliorate the model performance.

The convolutional neural network (CNN) has achieved extraordinary performance in image classification. However, the implementation of such architecture on embedded platforms is a big challenge task due to the computing resource constraint issue. This paper concentrates on optimization of CNN on embedded platforms with a case study of pedestrian detection in ADAS. The main contribution of this proposed CNN is its ability to run pedestrian classification task in real time with high accuracy based on a platform with ARM embedded. The CNN model has been trained with GPU locally and then transformed into an efficient implementation on embedded platforms. The efficient implementation uses dramatically small network scale and a lightweight CNN is obtained. Specifically, parameters of the network are compressed by adopting integer weights to reduce computational complexity. Meanwhile, other optimizations have also been proposed to adapt the general ARM processor architecture.