A Review of Multi-modal Human Motion Recognition Based on Deep Learning

Human motion recognition is a research hotspot in the field of computer vision,which has a wide range of applications,including biometrics,intelligent surveillance and human-computer interaction.In visionbased human motion recognition,the main input modes are RGB,depth image and bone data.Each mode can capture some kind of information,which is likely to be complementary to other modes,for example,some modes capture global information while others capture local details of an action.Intuitively speaking,the fusion of multiple modal data can improve the recognition accuracy.In addition,how to correctly model and utilize spatiotemporal information is one of the challenges facing human motion recognition.Aiming at the feature extraction methods involved in human action recognition tasks in video,this paper summarizes the traditional manual feature extraction methods from the aspects of global feature extraction and local feature extraction,and introduces the commonly used feature learning models of feature extraction methods based on deep learning in detail.This paper summarizes the opportunities and challenges in the field of motion recognition and looks forward to the possible research directions in the future.

Developmental toxicity of TCBPA on the nervous and cardiovascular systems of zebrafish (Danio rerio): A combination of transcriptomic and metabolomics

Tetrachlorobisphenol A(TCBPA),a widely used halogenated flame retardant,is frequently detected in environmental compartments and human samples.However,unknown developmental toxicity and mechanisms limit the entire understanding of its effects.In this study,zebrafish(Danio rerio)embryos were exposed to various concentrations of TCBPA while a combination of transcriptomics,behavioral and biochemical analyzes as well as metabolomics were applied to decipher its toxic effects and the potential mechanisms.We found that TCBPA could interfere with nervous and cardiovascular development through focal adhesion and extracellular matrix-receptor(ECM-receptor)interaction pathways through transcriptomic analysis.Behavioral and biochemical analysis results indicated abnormal swimming behavior of zebrafish larvae.Morphological observations revealed that TCBPA could cause the loss of head blood vessels.Metabolomic analysis showed that arginine-related metabolic pathways were one of the main pathways leading to TCBPA developmental toxicity.Our study demonstrated that by using omics,TCBPA was shown to have neurological and cardiovascular developmental toxicity and the underlying mechanisms were uncovered and major pathways identified.

Feather-like NiCo2O4 self-assemble from porous nanowires as binder-free electrodes for low charge transfer resistance

The unique feather-like arrays composing of ultrathin secondary nanowires are fabricated on nickel foam(NF)through a facile hydrothermal strategy.Thus,the enhancement of electrochemical properties especially the low charge transfer resistance strongly depends on more active sites and porosity of the morphology.Benefiting from the unique structure,the optimized NiCo2O4 electrode delivers a significantly lower charge transfer resistance of 0.32Ω and a high specific capacitance of450 Fg^-1 at 0.5 A.g^-1,as well as a superior cycling stability of 139.6%capacitance retention.The improvement of the electrochemical energy storage property proves the potential of the fabrication of various binary metal oxide electrodes for applications in the electrochemical energy field.

Responsive Trimodal Probes for In Vivo Imaging of Liver Inflammation by Coassembly and GSH-Driven Disassembly

Noninvasive in vivo imaging of hepatic glutathione(GSH)levels is essential to early diagnosis and prognosis of acute hepatitis.Although GSH-responsive fluorescence imaging probes have been reported for evaluation of hepatitis conditions,the low penetration depth of light in liver tissue has impeded reliable GSH visualization in the human liver.We present a liver-targeted and GSH-responsive trimodal probe(GdNPs-Gal)for rapid evaluation of lipopolysaccharide-(LPS-)induced acute liver inflammation via noninvasive,real-time in vivo imaging of hepatic GSH depletion.GdNPs-Gal are formed by molecular coassembly of a GSH-responsive Gd(III)-based MRI probe(1-Gd)and a liver-targeted probe(1-Gal)at a mole ratio of 5/1(1-Gd/1-Gal),which shows high r_(1) relaxivity with low fluorescence and fluorine magnetic resonance spectroscopic(^(19)F-MRS)signals.Upon interaction with GSH,1-Gd and 1-Gal are cleaved and GdNPs-Gal rapidly disassemble into small molecules 2-Gd,2-Gal,and 3,producing a substantial decline in r_(1) relaxivity with compensatory enhancements in fluorescence and ^(19)F-MRS.By combining in vivo magnetic resonance imaging(^(1)H-MRI)with ex vivo fluorescence imaging and ^(19)F-MRS analysis,GdNPs-Gal efficiently detect hepatic GSH using three independent modalities.We noninvasively visualized LPS-induced liver inflammation and longitudinally monitored its remediation in mice after treatment with an anti-inflammatory drug,dexamethasone(DEX).Findings highlight the potential of GdNPs-Gal for in vivo imaging of liver inflammation by integrating molecular coassembly with GSH-driven disassembly,which can be applied to other responsive molecular probes for improved in vivo imaging.