Search

Object detection, a fundamental task in computer vision, is crucial for various intelligent edge computing applications. However, object detection algorithms are usually heavy in computation, hindering their deployments on resource-constrained edge devices. Traditional edge-cloud collaboration schemes, like deep neural network (DNN) partitioning across edge and cloud, are unfit for object detection due to the significant communication costs incurred by the large size of intermediate results. To this end, we propose a Difficult-Case based Small-Big model (DCSB) framework. It employs a difficult-case discriminator on the edge device to control data transfer between the small model on the edge and the large model in the cloud. We also adopt regional sampling to further reduce the bandwidth consumption and create a discriminator zoo to accommodate the varying networking conditions. Additionally, we extend DCSB to video tasks by developing an adaptive sampling rate update algorithm, aiming to minimize computational demands without sacrificing detection accuracy. Extensive experiments show that DCSB can detect 97.26%-97.96% objects while saving 74.37%-82.23% network bandwidth, compared to cloud-only methods. Furthermore, DCSB significantly outperforms the latest DNN partitioning methods, reducing inference time by 92.60%-95.10% given an 8Mbps transmission bandwidth. In video tasks, DCSB matches the detection accuracy of leading video analysis methods while cutting the computational overhead by 40%.

Visual interpretation is extremely important in human geography as the primary technique for geographers to use photograph data in identifying, classifying, and quantifying geographic and topological objects or regions. However, it is also time‐consuming and requires overwhelming manual effort from professional geographers. This paper describes our interdisciplinary team's efforts in integrating computer vision models with geographers' visual image interpretation process to reduce their workload in interpreting images. Focusing on the dune segmentation task, we proposed an approach called DCV2I${\bf DCV}^2{\bf I}$featuring a deep dune segmentation model to identify dunes and label their ranges in an automated way. By developing a tool to connect our model with ArcGIS—one of the most popular workbenches for visual interpretation, geographers can further refine the automatically generated dune segmentation on images without learning any CV or deep learning techniques. Our approach thus realized a noninvasive change to geographers' visual interpretation routines, reducing their manual efforts while incurring minimal interruptions to their work routines and tools they are familiar with. Deployment with a leading Chinese geography research institution demonstrated the potential of DCV2I${\bf DCV}^2{\bf I}$in supporting geographers in researching and solving drylands desertification.

The increasing deployment of multiple deep neural networks (DNNs) on edge devices is revolutionizing mobile vision applications, spanning autonomous vehicles, augmented reality, and video surveillance. These applications demand adaptation to contextual and environmental drifts, typically through fine-tuning on edge devices without cloud access, due to increasing data privacy concerns and the urgency for timely responses. However, fine-tuning multiple DNNs on edge devices faces significant challenges due to the substantial computational workload. In this paper, we present PatchLine, a novel framework tailored for efficient on-device training in the form of fine-tuning for multi-DNN vision applications. At the core of PatchLine is an innovative lightweight adapter design called patches coupled with a strategic patch updating approach across models. Specifically, PatchLine adopts drift-adaptive incremental patching, correlation-aware warm patching, and entropy-based sample selection, to holistically reduce the number of trainable parameters, training epochs, and training samples. Experiments on four datasets, three vision tasks, four backbones, and two platforms demonstrate that PatchLine reduces the total computational cost by an average of 55% without sacrificing accuracy compared to the state-of-the-art.

Selective laser sintering (SLS) is a single-step, three-dimensional printing (3DP) process that is gaining momentum in the manufacturing of pharmaceutical dosage forms. It also offers opportunities for manufacturing various pharmaceutical dosage forms with a wide array of drug delivery systems. This research aimed to introduce carbonyl iron as a multifunctional magnetic and heat conductive ingredient for the fabrication of oral tablets containing isoniazid, a model antitubercular drug, via SLS 3DP process. Furthermore, the effects of magnetic iron particles on the drug release from the SLS printed tablets under a specially designed magnetic field was studied. Optimization of tablet quality was performed by adjusting SLS printing parameters. The independent factors studied were laser scanning speed, hatching space, and surface/chamber temperature. The responses measured were printed tablets’ weight, hardness, disintegration time, and dissolution performance. It has been observed that, for the drug formulation with carbonyl iron, due to its inherent thermal conductivity, sintering tablets required relatively lower laser energy input to form the tablets of the same quality attributes as the other batches that contained no magnetic particles. Also, printed tablets with carbonyl iron released 25% more drugs under a magnetic field than those without it. It can be claimed that magnetic nanoparticles appear as an alternative conductive material to facilitate the sintering process during SLS 3DP of dosage forms.

The pharmaceutical industry is progressing swiftly, driven by innovations that have led to the discovery of novel active pharmaceutical ingredients (APIs), along with new delivery platforms and technologies. Despite these advancements, traditional processing methods continue to encounter significant challenges in delivering new chemical entities. Many of these entities exhibit inherent issues such as poor solubility and permeability, which result in low bioavailability. Furthermore, there are constraints in customizing and delivering medications tailored to the pharmacokinetic variability of paediatric and geriatric patients, highlighting the need for advanced technological solutions in drug delivery. Additive manufacturing (AM), or 3D printing, represents one such technological breakthrough in the pharmaceutical sector. This review examines the latest advancements in 3D printing technology and its applications within the field. It focuses on the implementation of various 3D printing techniques to design and create dosage forms for multiple delivery routes, addressing the unique challenges posed by each route. In addition to exploring the research involved in developing novel formulations, the review also considers the regulatory hurdles that must be overcome to bring these dosage forms to market. It evaluates the market demand and financial considerations of this segment, assessing the feasibility of scaling up production to meet the requirements of the target patient population.

Aqueous zinc-ion batteries promise good commercial application prospects due to their environmental benignity and easy assembly under atmospheric conditions, positioning them as a viable alternative to lithium-ion batteries. However, some inherent issues, such as chaotic zinc dendrite growth and inevitable side reactions, challenge the commercialization progress. In this work, we imprint highly ordered zinc microwall arrays to regulate the electric field toward uniform Zn deposition. Afterward, coating a polyethylene glycol protection layer on the zinc microwalls aims to passivate the surface defects that rise unintentionally by mechanical imprinting. Polyethylene glycol can also boost oriented Zn deposition along the (002) plane and inhibit hydrogen gas production, further enhancing the stability of such three-dimensional (3D) hybrid anodes. Compared to the messy electric field near the polyethylene glycol-protected Zn foil, the uniform electric field provided by these 3D hybrid anodes can regulate the Zn deposition behaviors, enabling a longer lifespan and thus certifying the necessity of adding 3D microstructures. Additionally, 3D microstructures can offer a larger surface area than that of the planar Zn foil, providing more reaction sites and higher specific capacity. In this case, the 3D hybrid electrode exhibits a good initial capacity of approximately 120 mA h/g at a current density of 5 A/g and a nice retention of more than 80% after 800 cycles. The proposed scheme paves the way for a long-term stable 3D zinc anode solution with promising application prospects.

With increasing application demands of electronics and electric vehicles, the energy density of lithium-ion batteries (LIBs) is expected to be higher and higher. The silicon-based anode materials have triggered global research interest due to low operating voltage and high specific capacity. However, for the Si-based anode, the large volume change during cycling causes cracking and pulverization of Si particles, leading to the sluggish kinetics and poor cycle life. In this work, fluoroethylene carbonate (FEC) and lithium bis(fluorosulfonyl)imide (LiFSI) are used as synergistic functional additives to enhance the performance of silicon–carbon (SiC) composite anode in pouch cell. The properties of solid electrolyte interphase (SEI) formed on the surface of SiC composite anode have been systematically investigated. The images of different electrolytes infiltration and gas production after formation are analyzed with ultrasonic transmission scanning technique. DFT calculations are used to illustrate the mechanism. All date collection is at pouch cell level, which is more persuasive.

We report the case of a 55-year-old man who was transferred from an overseas institution with an initial impression of non-ST-elevation myocardial infarction (NSTEMI). Subsequent investigation found a severe aortic regurgitation (AR) complicated by acute ischemic hepatitis and acute renal injury, and the patient developed cardiopulmonary collapse with pulseless electrical activity perioperatively. Cardiopulmonary resuscitation was commenced immediately, venoarterial extracorporeal membrane oxygenation (ECMO) was immediately set up to bridge for emergency aortic-valve replacement surgery. The patient was subsequently weaned off ECMO and made a complete recovery. We demonstrate the successful use of ECMO for resuscitation in a patient with AR complicated by cardiovascular collapse.

Indirect laryngoscopy, in the form of flexible fibrescopy remains the gold standard in the management of predicted difficult intubation. In the last decade, indirect laryngoscopy in the form of videolaryngoscopy has quickly gained popularity as a tool in the anaesthetists' armamentarium in difficult airway management. We describe a case of successful intubation in a difficult airway with the combination of both videolaryngoscopy and flexible fibrescopy when attempts with other commonly used techniques failed.