Aiming at the problem that it is difficult to distinguish similar land types in outdoor scenes with multiple objects and complex spatial topological relationships, an A-Edge-SPG (Attention-EdgeConv SuperPoint Graph) graph neural network combining graph model and attention mechanism module was proposed. Firstly, the superpoints were segmented by the combination of graph cut and geometric features. Secondly, the local adjacency graph was constructed inside the superpoint to capture the context information of the point cloud in the scene and use the attention mechanism module to highlight the key information. Finally, a SuperPoint Graph (SPG) model was constructed, and the features of hyperpoints and hyperedges were aggregated by Gated Recurrent Unit (GRU) to realize accurate segmentation among different land types of point cloud. On Semantic3D dataset, the semantic segmentation effect of A-Edge-SPG model and SPG-Net (SPG neural Network) model was compared and analyzed. Experimental results show that compared with the SPG model, A-Edge-SPG model improves the Overall segmentation Accuracy(OA), mean Intersection over Union (mIoU) and mean Average Accuracy (mAA) by 1.8, 5.1 and 2.8 percentage points respectively, and significantly improves the segmentation accuracy of similar land types such as high vegetation and dwarf vegetation, improving the effect of distinguishing similar land types.
In the framework of traditional knowledge distillation, the teacher network transfers all of its own knowledge to the student network, and there are almost no researches on the transfer of partial knowledge or specific knowledge. Considering that the industrial field has the characteristics of single scene and small number of classifications, the evaluation of recognition performance of neural network models in specific categories need to be focused on. Based on the attention feature transfer distillation algorithm, three specific knowledge learning algorithms were proposed to improve the classification performance of student networks in specific categories. Firstly, the training dataset was filtered for specific classes to exclude other non-specific classes of training data. On this basis, other non-specific classes were treated as background and the background knowledge was suppressed in the distillation process, so as to further reduce the impact of other irrelevant knowledge on specific classes of knowledge. Finally, the network structure was changed, that is the background knowledge was suppressed only at the high-level of the network, and the learning of basic graphic features was retained at the bottom of the network. Experimental results show that the student network trained by a specific knowledge learning algorithm can be as good as or even has better classification performance than a teacher network whose parameter scale is six times of that of the student network in specific category classification.
In the field of traffic safety, the road abandoned objects easily cause traffic accidents and become potential traffic safety hazards. Focusing on the problems of low recognition rate and poor detection effect for different abandoned objects of traditional road abandoned object detection methods, a road abandoned object detection algorithm based on the optimized instance segmentation model CenterMask was proposed. Firstly, the residual network ResNet50 optimized by dilated convolution was used as the backbone neural network to extract image features and carry out the multi-scale processing. Then, the Fully Convolutional One-Stage (FCOS) target detector optimized by Distance Intersection over Union (DIoU) function was used to realize the detection and classification of road abandoned objects. Finally, the spatial attention-guided mask was used as the mask segmentation branch to realize the object shape segmentation, and the model training was realized by the transfer learning method. Experimental results show that, the detection rate of the proposed algorithm for road abandoned objects is 94.82%, and compared with the common instance segmentation algorithm Mask Region-Convolutional Neural Network (Mask R-CNN), the proposed road abandoned object detection algorithm has the Average Precision (AP) increased by 8.10 percentage points in bounding box detection.
Due to the lack of performance analysis while designing a distributed Evolutionary Algorithm (dEA), the designed algorithm cannot reach the expected speedup. To solve this problem, a comprehensive performance analysis method was proposed. According to the components of dEAs, factors that influence the performance of dEAs can be divided into three parts, namely, evolutionary cost, fitness evaluation cost and communication cost. Firstly, the feature of evolutionary cost under different individual encoding lengths was studied. Then when the evolutionary cost was kept unchanged, the fitness evaluation cost was controlled by using the delay function of the operating system and the communication cost was controlled by changing the length of individual encoding. Finally, the effect of each factor was tested through control variable method. The experimental results reveal the constraint relation among the three factors and point out the necessary conditions for speeding up dEAs.