Accurate and efficient traffic prediction is the key to the realization of intelligent transportation system (ITS), which helps to alleviate traffic congestion and reduce traffic accidents. Due to the complex dynamic spatial-temporal dependence between traffic networks, traffic prediction is extremely challenging. In previous studies, convolution neural network (CNN) and graph convolution network (GCN) were used to model spatial correlation. However, the non-Euclidean correlation of road network reduces the effect of convolution operator modeling. In addition, only considering the traffic interaction around the concerned points simplifies the influence of traffic network. In order to address the above problems, this article proposes an end-to-end global spatial-temporal graph attention network (GST-GAT), which uses the “global interaction + node query” to model the dynamic spatial-temporal correlation of traffic. In the encoder, the long short-term memory (LSTM) component flexibly transforms the traffic dynamic spatial-temporal graph into feedforward differentiable features. Global traffic interaction is proposed to summarize traffic network context changes and integrate all node features at each moment through a forward calculation. Then, each node computes the influence of traffic global interaction on a single node in parallel, and the spatial-temporal interaction information is adaptive fused by gating fusion mechanism. Finally, the end-to-end network structure is used to train the rich mixed feature coding to generate the traffic prediction status of each node. Experiments on public transportation data sets show that GST-GAT performs better than previous work in terms of accuracy and inference speed.

Edge agents, represented by socially-aware robots and autonomous vehicles, have gradually been integrated into human society. The safety navigation system in interactive scenes is of great importance to them. The key of this system is that the edge agent has the ability to predict the pedestrian trajectory in the dynamic scene, so as to avoid collision. However, predicting pedestrian trajectories in dynamic scenes is not an easy task, because it is necessary to comprehensively consider the spatial-temporal structure of human-environment interaction, visual attention, and the multi-modal behavior of human walking. In this paper, a scalable spatial-temporal graph generation adversarial network architecture (STG-GAN) is introduced, which can comprehensively consider the influence of human-environment interaction and generate a reasonable multi-modal prediction trajectory. First, we use LSTM nodes to flexibly transform the spatial-temporal graph of human-environment interactions into feed-forward differentiable feature coding, and innovatively propose the global node to integrate scene context information. Then, we capture the relative importance of global interactions on pedestrian trajectories through scaled dot product attention, and use recurrent sequence modeling and generative adversarial network architecture for common training, so as to generate reasonable pedestrian future trajectory distributions based on rich mixed features. Experiments on public data sets show that STG-GAN is superior to previous work in terms of accuracy, reasoning speed and rationality of trajectory prediction.

Deep learning has become popular in recent years primarily due to the powerful computing device such as GPUs. However, deploying these deep models to end-user devices, smart phones, or embedded systems with limited resources is challenging. To reduce the computation and memory costs, we propose a novel lightweight deep learning module by low-rank pointwise residual (LPR) convolution, called LPRNet. Essentially, LPR aims at using low-rank approximation in pointwise convolution to further reduce the module size, while keeping depthwise convolutions as the residual module to rectify the LPR module. This is critical when the low-rankness undermines the convolution process. We embody our design by replacing modules of identical input-output dimension in MobileNet and ShuffleNetv2. Experiments on visual recognition tasks including image classification and face alignment on popular benchmarks show that our LPRNet achieves competitive performance but with significant reduction of Flops and memory cost compared to the state-of-the-art deep models focusing on model compression.