Multi-task scene semantic comprehension model based on novel neural network, and application thereof

Abstract

The invention provides a lightweight and efficient neural network model, and designs a multi-task scene semantic comprehension model based on the model. The neural network comprises the following fivestages that: stage one: adopting an mC.ReLU module; stage two: adopting Max-Pooling to carry out downsampling, and then, adopting a plurality of Block-B modules; stage three: adopting a plurality ofBlock-B modules and a plurality of Block-D modules; stage four: adopting a plurality of Block-A modules and a plurality of Block-C modules; and stage five: adopting a plurality of Block-A modules. Themulti-task scene semantic comprehension model comprises the neural network model, a multilevel fusion network and a multi-task semantic prediction network, wherein the neural network model and the multilevel fusion network share different tasks to serve as a common feature extractor. The model can be applied to the field of automatic driving scene perception, and is favorable for realizing road obstacle detection, scene semantic segmentation and real-time multi-task semantic prediction.

Description

technical field [0001] The invention relates to the field of scene semantic understanding, in particular to a novel neural network-based multi-task scene semantic understanding model and its application in an automatic driving system. Background technique [0002] Autonomous driving is an important application area for 3D scene understanding. The 3D scene understanding technology is applied in the automatic driving system, which can realize tasks such as object detection and scene semantic segmentation. Visual recognition models based on neural networks have been proven to have excellent performance in a large number of existing works. However, applying neural networks to visual perception computing on mobile platforms is still challenging. [0003] On the one hand, due to the high computational complexity of neural networks, it is difficult for popular VGG, GoogleNet, ResNet-50 and other networks to achieve real-time calculations on embedded devices; on the other hand, th...

AI Technical Summary

Problems solved by technology

[0004] However, in an actual autonomous driving system, due to limited hardware computing resources,

Images