As an essential part of Internet of Things, monocular depth estimation predicts dense depth maps from a single RGB image captured by monocular cameras. Past monocular depth estimation methods almost focus on improving accuracy at the cost of increased latency, power consumption, and computational complexity, failing to balance accuracy and efficiency. Additionally, when speeding up depth estimation algorithms, researchers commonly ignore their adaptation to different hardware architectures on edge devices. This paper aims to solve these challenges. (1) We design an efficient monocular depth estimation model (MDE) for precise depth sensing on edge devices. (2) We employ a reinforcement learning algorithm and automatically prune redundant channels of MDE by finding a relatively optimal pruning policy. The pruning approach lowers model runtime and power consumption with little loss of accuracy through achieving a target pruning ratio. (3) We accelerate the pruned MDE, while adapting it to different hardware architectures with a compilation optimization method. The compilation optimization further reduces model runtime by an order of magnitude on hardware architectures. Extensive experiments confirm that our methods are effective for images of different sizes on two public datasets. The pruned and optimized MDE achieves promising depth sensing with a better trade-off among model runtime, accuracy, computational complexity, and power consumption than the state of the arts on different hardware architectures.

High-speed railways have been one of the most popular means of transportation all over the world. As an important part of the high-speed railway power supply system, the overhead catenary system (OCS) directly influences the stable operation of the railway, so regular inspection and maintenance are essential. Now manual inspection is too inefficient and high-cost to fit the requirements for high-speed railway operation, and automatic inspection becomes a trend. The 3D information in the point cloud is useful for geometric parameter measurement in the catenary inspection. Thus it is significant to recognize the components of OCS from the point cloud data collected by the inspection equipment, which promotes the automation of parameter measurement. In this paper, we present a novel method based on deep learning to recognize point clouds of OCS components. The method identifies the context of each single frame point cloud by a convolutional neural network (CNN) and combines some single frame data based on classification results, then inputs them into a segmentation network to identify OCS components. To verify the method, we build a point cloud dataset of OCS components that contains eight categories. The experimental results demonstrate that the proposed method can detect OCS components with high accuracy. Our work can be applied to the real OCS components detection and has great practical significance for OCS automatic inspection.

Given monocular images as inputs, monocular depth estimation (MDE) infers pixel-level depth. MDE is always a critical stage in scene sensing on edge devices. Existing MDE studies frequently employ deep neural networks (DNNs) for MDE, but they still face some problems, such as sacrificing computational complexity and efficiency in return for great precision, or losing more precision in exchange for increased efficiency. To alleviate these issues; 1) we propose an encoder–decoder network (EdgeNet) for precise and fast MDE on different edge devices. When recovering depth in the decoder, we design upsampling modules to aggregate global depth information with low computational complexity, improving the accuracy of the decoder by extracting its different ranges of depth information; 2) we develop a two-stage channel pruning method to, respectively, prune the encoder and decoder based on their characteristics. Our pruning method further reduces latency and model/computational complexity of EdgeNet, while losing little accuracy; and 3) we optimize the pruned EdgeNet to decrease graphics processing unit (GPU) scheduling overhead. The optimization accelerates MDE inference by an order of magnitude on the TX2 GPU device, when the input resolution is 224 $\times $ 224. Extensive experiments show that our strategies are effective on different edge GPU devices, when input resolutions differ in outdoor or indoor scenes. For example, compared with the state of the art, the optimized EdgeNet, respectively, reduces the GPU latency by 76.3% and 89.2% on Nano and TX2 GPU devices with 2.6% lower root mean square error when the input resolution is 128 $\times $ 416.

As overhead contact (OC) is an essential part of power supply systems in high-speed railways, it is necessary to regularly inspect and repair abnormal OC components. Relative to manual inspection, applying LiDAR (light detection and ranging) to OC inspection can improve efficiency, accuracy, and safety, but it faces challenges to efficiently and effectively segment LiDAR point cloud data and identify catenary components. Recent deep learning-based recognition methods are rarely employed to recognize OC components, because they have high computational complexity, while their accuracy needs to be improved. To track these problems, we first propose a lightweight model, RobotNet, with depthwise and pointwise convolutions and an attention module to recognize the point cloud. Second, we optimize RobotNet to accelerate its recognition speed on embedded devices using an existing compilation tool. Third, we design software to facilitate the visualization of point cloud data. Our software can not only display a large amount of point cloud data, but also visualize the details of OC components. Extensive experiments demonstrate that RobotNet recognizes OC components more accurately and efficiently than others. The inference speed of the optimized RobotNet increases by an order of magnitude. RobotNet has lower computational complexity than other studies. The visualization results also show that our recognition method is effective.

The overhead contact system (OCS) is a critical railway infrastructure for train power supply. Periodic inspections, aiming at acquiring the operational condition of the OCS and detecting problems, are necessary to guarantee the safety of railway operations. One of the OCS inspection means is to analyze data of point clouds collected by mobile 2D LiDAR. Recognizing OCS components from the collected point clouds is a critical task of the data analysis. However, the complex composition of OCS makes the task difficult. To solve the problem of recognizing multiple OCS components, we propose a new deep learning-based method to conduct semantic segmentation on the point cloud collected by mobile 2D LiDAR. Both online data processing and batch data processing are supported because our method is designed to classify points into meaningful categories of objects scan line by scan line. Local features are important for the success of point cloud semantic segmentation. Thus, we design an iterative point partitioning algorithm and a module named as Spatial Fusion Network, which are two critical components of our method for multi-scale local feature extraction. We evaluate our method on point clouds where sixteen categories of common OCS components have been manually labeled. Experimental results show that our method is effective in multiple object recognition since mean Intersection-over-Unions (mIoUs) of online data processing and batch data processing are, respectively, 96.12% and 97.17%.

Traffic sign detection is one of the critical technologies in the field of intelligent transportation systems (ITS). The difficulty of traffic sign detection mainly lies in detecting small objects in a wide and complex traffic scene quickly and accurately. In this paper, we regard traffic sign detection as a region classification problem and propose a two-stage CNN-based approach to solve it. At the first stage, we design an efficient network which is built with improved fire-modules to generate object proposals quickly. The network up-samples and merges the feature maps of different scales to attain a high-resolution fused feature map which contains semantically strong features of multi-scale objects. Specially, the prediction is made on the fuse feature map and based on the novel center-point estimation. With the overall designs, our region proposal network can achieve high recall value while using low-resolution images. At the second stage, a separate classification network is proposed. The bottleneck of the classification performance is generally caused by the greatly similar appearances between traffic signs. Therefore, we further explore local regions with critical differences between traffic signs to obtain fine-grained local features which help to improve classification. Finally, we evaluate our method on a challenge benchmark Tsinghua-Tencent 100K which provides many large images with small traffic sign instances. The experiment result shows that our method has better performance and faster detection speed than many state-of-the-art traffic sign detection methods.