In recent years, the convolutional neural network (CNN) has made remarkable achievements in semantic segmentation. The\nmethod of semantic segmentation has a desirable application prospect. Nowadays, the methods mostly use an encoder-decoder\narchitecture as a way of generating pixel by pixel segmentation prediction. The encoder is for extracting feature maps and\ndecoder for recovering feature map resolution. An improved semantic segmentation method on the basis of the encoderdecoder\narchitecture is proposed. We can get better segmentation accuracy on several hard classes and reduce the computational\ncomplexity significantly. This is possible by modifying the backbone and some refining techniques. Finally, after some\nprocessing, the framework has achieved good performance in many datasets. In comparison with the traditional architecture,\nour architecture does not need additional decoding layer and further reuses the encoder weight, thus reducing the complete\nquantity of parameters needed for processing. In this paper, a modified focal loss function is also put forward, as a replacement\nfor the cross-entropy function to achieve a better treatment of the imbalance problem of the training data. In addition, more\ncontext information is added to the decode module as a way of improving the segmentation results. Experiments prove that the\npresented method can get better segmentation results. As an integral part of a smart city, multimedia information plays an\nimportant role. Semantic segmentation is an important basic technology for building a smart city
Loading....