The deterministic and real-time communication is the indispensable requirement in Industrial Internet of Things (IIoT) application areas. Time-sensitive networking (TSN) a promising technology for this kind demands through designing proper scheduling routing mechanisms. However, it still challenging to design mechanisms large-scale instances due high computational complexity. In order guarantee schedulability scalability, learning-based scalable co-design (LSSR) architecture proposed article...

The attention mechanism has become the de facto module in scene text recognition (STR) methods, due to its capability of extracting character-level representations. These methods can be summarized into implicit based and supervised based, depended on how is computed, i.e., are learned from sequence-level annotations or bounding box annotations, respectively. Implicit attention, as it may extract coarse even incorrect spatial regions character prone suffering an alignment-drifted issue....

Detecting the marking characters of industrial metal parts remains challenging due to low visual contrast, uneven illumination, corroded character structures, and cluttered background part images. Affected by these factors, bounding boxes generated most existing methods locate low-contrast text areas inaccurately. In this paper, we propose a refined feature-attentive network (RFN) solve inaccurate localization problem. Specifically, design parallel feature integration mechanism construct an...

The demands of reliable and real-time communication in industrial automation systems drive growing attention on Time-Sensitive Networks (TSNs) due to its guarantee low latency deterministic transmission. Current works explore iterated scheduling time-triggered (TT) streams TSNs mainly by random stream partitioning graph-based partitioning. However, obtains limited performance, while the edge weights heavily depend prior domain knowledge. In this article, we propose a...

360 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula> video becomes increasingly attractive in smart factories due to its immersive experience for industrial surveillance. However, transmitting this kind of requires a fairly large demand on the bandwidth high-resolution and panoramic view. Moreover, low-latency responses users' head movements are required. This leads...

Nighttime semantic segmentation has attracted considerable attention due to its crucial status in the smart city. However, it is challenging handle poor illumination and indiscernible information. To tackle these problems, a saliency-guided domain adaptation network, SGDA, proposed via adapting daytime models nighttime scenes. Firstly, saliency guidance branch attached network enrich spatial features guide model better perceive detail Secondly, embed pyramid architecture designed fuse from...

Electric trucks (ETs) possess large-capacity batteries and are widely adopted in express industry. The rapid development of battery technologies discharging facilities endows ETs with another role as temporary mobile energy storage (MES) to deliver feed back grid at charging stations. Thus, widespread have great potential relieve load pressure on transmission lines the assist allocate energy. By assuming that couriers (ECs, ETs) can be recruited through cloud platform, surplus renewable (RE)...

The inverter is the most vulnerable module of photovoltaic (PV) systems. insulated gate bipolar transistor (IGBT) core part inverters and root source PV failures. How to effectively diagnose IGBT faults critical for reliability, high efficiency, safety Recently, deep learning (DL) methods are widely used fault detection diagnosis. Different from traditional diagnosis methods, DL use neural networks which can automatically extract useful representative features raw data. However, require...

Realtime and intelligent video surveillance via camera networks involve computation-intensive vision detection tasks with massive data, which is crucial for safety in the edge-enabled industrial Internet of Things (IIoT). Multiple streams compete limited communication resources on link between edge devices networks, resulting considerable congestion. It postpones completion time degrades accuracy tasks. Thus, achieving high under constraints task deadline challenging. Previous works focus...

Crowdsourcing is a relatively economic and efficient solution to collect annotations from the crowd through online platforms. Answers collected workers with different expertise may be noisy unreliable, quality of annotated data needs further maintained. Various solutions have been attempted obtain high-quality annotations. However, they all assume that workers' label stable over time (always at same level whenever conduct tasks). In practice, attention changes time, ignorance which can...

Object detection is crucial for surveillance in edge-enabled Industrial Internet-of-Things. Massive high-dimensional video streams without considering priority differences connect to edges via narrow and time-varying uplink channels, which should be analyzed efficiently accurate fast responses. However, network environments constrained edge resources degrade surveillance's accuracy real-time performance. This article proposes EdgeLeague multiple with different quality of service, maintains...

Multiobject tracking (MOT) is a crucial technology for security surveillance, which computationally intensive due to the requirement of processing large number video streams within low latency in practice. The input MOT are processed on cloud computing center with abundant computational capability, posing heavy pressures delivering cloud. Recent advances Internet-of-Things (IoT) provide edge-computing-based solutions analytics at scale. However, gap between MOT's high capability demand and...

The trajectory planning of autonomous vehicles requires making safe sequential decisions instantaneously. most significant challenge is the uncertainties brought by complex interactions with other road users in diverse driving scenarios. Previous rule-based solutions lack generalization and only suit for limited simple environments. Moreover, another approach generates feasible action sequences based on motion planning, which ignores between vehicle participants. This paper proposes a...

Multi-target multi-camera tracking (MTMCT) is a crucial component in traffic flow analysis for smart transportation systems. MTMCT generates vehicle trajectories from the surveillance videos across cameras at different intersections. Variable orientations during driving process degrade precision. Besides, time-varying between could lead to trajectory mismatching under constant travel time constraint algorithms. In this paper, an orientation-based method considering proposed. First, stacked...

The attention mechanism has become the \emph{de facto} module in scene text recognition (STR) methods, due to its capability of extracting character-level representations. These methods can be summarized into implicit based and supervised based, depended on how is computed, i.e., are learned from sequence-level annotations or bounding box annotations, respectively. Implicit attention, as it may extract coarse even incorrect spatial regions character prone suffering an alignment-drifted...

Scene text recognition (STR) methods combined with semantic information have made great progress to recognize texts in natural scenes, most of which are daily words. However, research on mining industrial attracts less attention. Since follow a different pattern defined by industry standards, it challenges many existing conduct accurate reasoning. In this paper, we abstract the standards into two aspects prior knowledge: grouping property and lexicon. Correspondingly, knowledge-based...

Realtime and intelligent video surveillance via camera networks involve computation-intensive vision detection tasks with massive data, which is crucial for safety in the edge-enabled industrial Internet of Things (IIoT). Multiple streams compete limited communication resources on link between edge devices networks, resulting considerable congestion. It postpones completion time degrades accuracy tasks. Thus, achieving high under constraints task deadline challenging. Previous works focus...