Real-time semantic segmentation method based on sequence knowledge distillation

Abstract

The invention discloses a real-time semantic segmentation method based on sequential knowledge distillation. The real-time semantic segmentation method comprises the following steps: step 1, acquiringan outdoor streetscape image; step 2, sending the acquired outdoor streetscape image to a sequence prediction network to obtain a semantic segmentation result, wherein the sequence prediction networkadopts an Xception network as a backbone network and is used for extracting image features, a codec in the sequence prediction network comprises a plurality of coding modules and a plurality of decoding modules, each module outputs a prediction result for an input image, and the prediction result of the previous module is used as a part of the prediction result input of the next module, so that the existing network is fully utilized, the network performance is improved, the parameters of the forward network are optimized during gradient updating, and the previous prediction result is optimized. According to the semantic segmentation method, information loss caused in the network transmission process can be fully reduced while only a small amount of calculation amount is increased, so thatthe network precision is further improved.

Description

technical field [0001] The invention belongs to the technical field of image segmentation, in particular to a real-time semantic segmentation method based on sequence knowledge distillation. Background technique [0002] In recent years, although deep learning has made great progress in the field of semantic segmentation, such as Long J, Shelhamer E, Darrell T.Fully convolutional networks for semantic segmentation[C] / / Proceedings of the IEEE conference on computer vision and pattern recognition.2015 :3431-3440., Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C] / / International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, 2015:234-241., Badrinarayanan V, Kendall A, Cipolla R. Segnet: A deep convolutional encoder-decoder architecture for imagesegmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39(12): 2481-2495., Chen L C, Papandreou G, Kokkinos I ...

AI Technical Summary

Problems solved by technology

In edge detection, it has been proposed to add supervision information in the encode stage. However, these methods directly upsample the prediction results of each module to the original resolution size, and then output the final prediction results through the convolutional layer after cascading. This method Not effective for semantic segmentation and extremely time consuming

[0004] The existing knowledge distillation method only distills the knowledge information of the Teacher network or supervised image on the last prediction result, and fails to fully exploit the potential of the network.

Images