PMID int64 35.7M 36.4M | Title/Abstract stringlengths 22 8.78k | MeshTerms stringlengths 7 101 | SemanticTypes stringlengths 4 64 |
|---|---|---|---|
36,236,603 | imped analysi viscoelast petal structur stearic acid function layer deposit qcm function layer crucial qcm sensor immobil target molecul microstructur layer determin sensit sensor hand microstructur also affect load sensor studi imped analysi use investig relationship microstructur viscoelast properti petal stearic acid sa function layer sa layer deposit use vacuum thermal evapor techniqu differ petal pillar structur elast layer generat vari deposit time analysi show growth embed pillar structur dramat reduc conduct increas bandwidth energi dissip vibrat could relat interact pillar elast matrix | D013229 | T109 |
36,236,601 | use smartphon locat trap victim disast larg unaccount number victim disast event fire often trap build debri must locat rescu soon possibl studi transform smartphon indoor locat tool without extra modif complic program instal consid smartphon like carri disast strike studi creat system convert smartphon lifesav tool trap victim rescuer studi employ bluetooth beacon smartphon send signal use low power consumpt featur signal could continu transmit rescuer locat trap victim longer rescuer could use bluetooth function regular notebook comput search signal without hardwar implement modif allow locat determin posit mani trap victim simultan implement system decreas search rescu team need enter unsaf area increas rescu speed critic factor surviv trap victim furthermor disast strike smartphon call function might work trap victim might weak call help thus autorepli messag victim smartphon could help locat within error even cover fallen debri wood pile tile effort increas chanc find trap victim within golden rescu hour reduc exposur time search rescu team unsaf environ | D004190;D018635 | T057;T067 |
36,236,600 | real time adapt traffic signal control connect autom vehicl environ optimis signal plan reinforc learn vehicl speed guidanc adapt traffic signal control atsc effect method reduc traffic congest modern urban area mani studi adopt various approach adjust traffic signal plan accord real time traffic respons demand fluctuat improv urban network perform eg minimis delay recent learn base method reinforc learn rl achiev promis result signal plan optimis howev adopt self learn techniqu futur traffic environ presenc connect autom vehicl cav remain larg open challeng studi develop real time rl base adapt traffic signal control optimis signal plan minimis total queue length allow cav adjust speed base fix time strategi decreas total stop delay highlight work combin speed guidanc system reinforc learn base traffic signal control two differ perform measur implement minimis total queue length total stop delay result indic propos method outperform fix time plan optim speed advisori cav environ tradit actuat control term averag stop delay vehicl queue length particular satur oversatur condit | D001334 | T056 |
36,236,598 | caregiv profil base canadian occup perform measur adopt assist technolog copm canadian occup perform measur tool base identif self perceiv perform satisfact problem perform occup allow creation hierarchi order intervent carri speed identif necessari assist technolog given import caregiv percept perform design research examin caregiv profil copm sampl caregiv volunt particip studi cluster analysi carri copm score two caregiv profil found relat copm measur one low score perform satisfact anoth high score two variabl main predictor found self percept perform structur replic hierarch cluster analysi role caregiv interest result relev theoret practic level | D009788;D012656 | T061;T074 |
36,236,597 | deep learn chondrogen tumor classif wavelet transform raman spectra grade cancer tissu still one main challeng pathologist develop enhanc analysi strategi henc becom crucial accur identifi deal individu case raman spectroscopi rs promis tool classif tumor tissu allow us obtain biochem map tissu analysi observ evolut term biomolecul protein lipid structur dna vitamin howev potenti could improv provid classif system would abl recogn sampl tumor categori take input raw raman spectroscopi signal could provid reliabl respons shorter time scale could reduc elimin fals posit negat diagnos deep learn techniqu becom ubiquit recent year model abl perform classif high accuraci divers field research eg natur languag process comput vision medic imag howev deep model often reli huge label dataset produc reason accuraci otherwis occur overfit issu train data insuffici paper propos chondrogen tumor classif wavelet transform raman spectra clara abl classifi high accuraci raman spectra obtain bone tissu clara recogn grade tumor evalu dataset accuraci exploit classif pipelin consist divis origin task two binari classif step first perform origin rs signal latter accomplish use hybrid tempor frequenc transform | D000077321;D009369 | T191;T066 |
36,236,596 | analysi phenol compound food coulometr array detector review phenol compound import group organ molecul high radic scaveng antimicrobi anti inflammatori antioxid properti emerg interest phenol compound food product led develop various analyt techniqu detect character among coulometr array detector sensit select precis method analysi polyphenol review discuss principl method recent advanc develop well trend applic analysi phenol compound food product fruit cereal beverag herb spice | D000890;D000975 | T121;T044;T123;T039 |
36,236,595 | target enclos coverag control quadrotor constraint time vari delay neural adapt fault toler format control approach paper investig problem format fault toler control multipl quadrotor qrs mobil sens orient applic qrs subject fault input satur time vari delay control perform target enclos cover task guarante state constraint exceed distribut format control scheme propos use radial basi function neural network rbfnn base time delay posit control adapt fault toler attitud control lyapunov krasovskii approach use analyz time vari delay barrier lyapunov function deploy handl prescrib constraint auxiliari system combin command filter design resolv satur problem rbfnn adapt estim deploy provid estim disturb fault signal uncertainti proven close loop signal bound propos protocol prescrib constraint violat enhanc flight safeti qr format applic compar simul base applic scenario verifi effect propos method | D016571;D017711 | T170;T077 |
36,236,592 | pedestrian trajectori predict real time autonom system via context augment transform network forecast trajectori pedestrian share urban traffic environ non invas sensor modal still consid one challeng problem face develop autonom vehicl av literatur problem often tackl use recurr neural network rnns despit power capabl rnns captur tempor depend pedestrian motion trajectori argu challeng deal longer sequenti data addit whilst accommod contextu inform scene semant agent interact shown effect robust trajectori predict also impact overal real time perform predict system thus work introduc framework base transform network demonstr recent effici outperform rnns mani sequenti base task reli fusion sensor modal name past posit inform agent interact inform scene physic semant inform input framework order provid robust trajectori predict pedestrian also achiev real time perform multi pedestrian trajectori predict evalu framework three real life dataset pedestrian share urban traffic environ outperform compar baselin approach short term long term predict horizon short term predict horizon approach achiev lower score accord averag displac error root mean squar error aderms predict state art sota approach cm cm respect long term predict horizon approach achiev lower ade fde sota approach cm cm respect addit approach achiev superior real time perform score ie provid individu trajectori predict per second | D000069636 | T098 |
36,236,591 | night vision anti halat method base infrar visibl video fusion order address discontinu caus direct applic infrar visibl imag fusion anti halat method video effici night vision anti halat method base video fusion propos design frame select base inter frame differ determin optim cosin angl threshold analyz relat cosin angl threshold nonlinear correl inform entropi de frame rate propos time mark base adapt motion compens construct number interpol frame redund frame take retain frame number time stamp time consid motion vector two adjac retain frame benchmark adapt weight construct accord interfram differ interpol frame last retain frame motion vector interpol frame estim experiment result show propos frame select strategi ensur maximum safe frame remov premis continu video content differ vehicl speed various halat scene frame number play durat fuse video consist origin video content interpol frame high synchron correspond origin frame averag fps video fusion work six time frame frame fusion effect improv anti halat process effici video fusion | D007089;D007090 | T062;T066;T060 |
36,236,590 | research influenc coil lc parallel reson detect effect induct wear debri sensor coil structur induct wear debri sensor play signific role effect wear debri detect accord characterist lc parallel reson capacitor coil connect parallel make sensor coil lc parallel reson state benefici improv abil detect wear particl paper mathemat model output induc electromot detect coil establish analyz influenc structur detect sensit enhanc sensor current rate chang disturb magnet field essenti resist nois interfer base coher demodul principl ad lock amplifi appli test platform amplifi identifi weak signal addit experi design test output signal debri condit differ origin output voltag sensor parallel structur meanwhil near reson state detect coil lc parallel circuit test output signal inform result show sensor detect sensit effect improv lc parallel coil structur sensor structur paramet design paper optim raw output amplif voltag abras particl detect v detect perform ferromagnet particl non ferromagnet particl test condit realiz detect abil ferromagnet abras particl non ferromagnet abras particl | D000669 | T073 |
36,236,586 | signific featur human activ recognit use tri axial acceleromet activ recognit use wearabl sensor becom essenti varieti applic tri axial acceleromet wide use sensor activ recognit although various featur use captur pattern classifi acceleromet signal recognis activ consensus best featur choos reduc number featur reduc comput cost complex enhanc perform classifi paper identifi signal featur signific discrimin power differ human activ also investig effect sensor placement locat sampl frequenc activ complex select featur comprehens list signal featur extract acceleromet signal four public avail dataset includ featur never use activ recognit featur signific measur use joint mutual inform maximis jmim method common signific featur among dataset identifi result show sensor placement locat signific affect recognit perform affect signific sub set featur result also show high sampl frequenc featur relat signal repeat regular show high discrimin power | D061725;D006802 | T060;T052 |
36,236,584 | kidney cancer predict empow blockchain secur use transfer learn kidney cancer danger lethal cancer diseas caus kidney tumor genet renal diseas patient surviv method earli predict kidney cancer earli predict kidney cancer help doctor start proper therapi treatment patient prevent kidney tumor renal transplant adapt artifici intellig autom tool empow differ deep learn machin learn algorithm predict cancer studi propos model use internet medic thing iomt base transfer learn techniqu differ deep learn algorithm predict kidney cancer earli stage patient data secur propos model incorpor blockchain technolog base privat cloud transfer learn train model predict kidney cancer propos model use biopsi cancer kidney consist three class propos model achiev highest train accuraci predict accuraci respect empow data augment without augment propos model achiev predict accuraci valid transfer learn provid promis framework combin iomt technolog blockchain technolog layer enhanc diagnos capabl kidney cancer | D000081186;D007680 | T191;T066 |
36,236,583 | automat modul recognit base optim linear combin higher order cumul automat modul recognit amr use various domain general purpos communic mani militari applic thank grow popular internet thing iot relat communic technolog research articl propos innov idea combin classic mathemat techniqu comput linear combin lcs cumul genet algorithm ga creat super cumul super cumul use classifi five digit modul scheme fade channel use k nearest neighbor knn propos classifi signific improv percentag recognit accuraci lower snrs use smaller sampl size comparison exist techniqu manifest supremaci propos classifi | D000465 | T170 |
36,236,581 | model trajectori obtain extern sensor locat predict via nlp approach represent learn seek extract use low dimension attribut complex high dimension data natur languag process nlp use investig represent learn model extract word featur vector use sequenti order text via word embed languag model maintain semant mean inspir nlp paper tackl represent learn problem trajectori use nlp method encod extern sensor posit road network generat featur space predict next vehicl movement evalu vector represent road sensor trajectori use extrins intrins strategi result shown potenti natur languag model describ space featur trajectori applic next locat predict | D009323;D012660 | T078;T066 |
36,236,580 | smartphon sensor base human motion character neural stochast differenti equat transform model mani conveni afford advanc smartphon technolog develop advanc data analysi method health relat inform smartphon user becom fast grow research topic healthcar field along line paper address smartphon sensor base character human motion neural stochast differenti equat nsdes transform model nsdes model via transform network two promin deep learn base model approach signific perform yield mani applic problem model dynam featur stochast differenti equat deep neural network frequent use paradigm scienc engin respect combin two paradigm one unifi framework drawn signific interest deep learn communiti nsdes among lead technolog combin effort use attent also becom wide adopt strategi mani deep learn applic transform deep learn model use mechan self attent concept self attent base transform origin introduc task natur languag process nlp due excel perform versatil scope applic rapid expand util techniqu neural stochast differenti equat transform model along data obtain smartphon sensor present deep learn method capabl effici character human motion character human motion encod high dimension sequenti data smartphon sensor latent variabl low dimension latent space concept latent variabl particular use carri condens inform concern motion data also learn low dimension represent precis use neural stochast differenti equat model transit human motion latent space reli generat pre train transform gpt base transform model approxim intract posterior condit latent variabl experi show propos method yield promis result problem character human motion pattern relat task includ user identif | D016571;D000068997 | T170;T073 |
36,236,579 | iotmapp metric aggreg system architectur support smart citi solut smart citi nowaday unavoid grow realiti support softwar platform support citi manag process present larg number data obtain sensor use throughout citi low power wide area network lpwan leverag sensor process howev urban landscap turn induc high probabl chang propag condit lpwan network thus requir activ monitor solut assess citi lpwan network condit current exist solut usual consid exist one type lpwan network monitor paper architectur aggreg metric heterogen lpwan network present architectur name iotmapp combin purpos build compon exist compon fiwar apach kafka ecosystem implement detail lpwan network abstract adapt new network may easili ad valid carri use real data collect long rang wide area network lorawan lisbon simul data set extrapol collect data result indic present architectur viabl solut metric aggreg may expand support multipl network howev consid fiwar compon present perform bottleneck may hinder scale architectur process new messag arriv | D019985;D017753 | T082;T081;T070 |
36,236,577 | character workspac limit oper laser treatment vascular lesion lower limb increas age popul bring numer challeng health aesthet segment use laser therapi dermatolog expect increas sinc allow non invas infect free treatment howev exist laser devic requir doctor manual handl visual inspect skin treatment outcom depend user expertis frequent result ineffect treatment side effect studi aim determin workspac limit oper laser treatment vascular lesion lower limb result studi use develop robot guid technolog help address aforement problem specif workspac limit oper studi eight vascular laser treatment electromagnet track system use collect real time posit laser treatment comput averag workspac length height width respect correspond averag volum treatment averag treatment time min averag laser orient degre addit averag veloc ms deg measur knowledg character vascular laser treatment workspac limit oper may eas understand futur robot system develop | D012371 | T090 |
36,236,575 | pedestrian flow predict rout recommend busi event due potenti econom benefit pedestrian flow consid essenti indic public space pedestrian flow predict design assist oper make decis shop center owner oper hold certain event sale promot attract surround pedestrian refer type event busi event busi event attract pedestrian flow mean busi opportun merchant moreov placement affect distribut pedestrian flow howev decid rout chosen specifi event difficult best knowledg first consid busi event predict pedestrian flow paper investig two problem one pedestrian flow predict busi event rout recommend busi event first propos attract base matrix factor model abmf effici predict pedestrian flow busi event introduc attract index differ categori pedestrian matrix factor second leverag skip gram mode learn latent represent improv pair wise rank loss flow awar base method sg fwarp aim learn event latent represent rout recommend compar state art method experiment result show abmf predict pedestrian flow matrix similar compar ground truth sg fwarp recommend rout busi event high accuraci | D000069636 | T098 |
36,236,574 | research ground object classif method high resolut remot sens imag base improv deeplabv ground object classif use remot sens imag high resolut wide use land plan ecolog monitor resourc protect tradit imag segment technolog poor effect complex scene high resolut remot sens imag field deep learn deep neural network appli high resolut remot sens imag segment deeplabv network deep neural network base encod decod architectur common use segment imag high precis howev segment accuraci high resolut remot sens imag poor number network paramet larg cost train network high therefor paper improv deeplabv network first mobilenetv network use backbon featur extract network attent mechan modul ad featur extract modul aspp modul introduc focal loss balanc design follow advantag enhanc abil network extract imag featur reduc network train cost achiev better semant segment accuraci experi high resolut remot sens imag dataset show miou propos method whdld dataset higher tradit deeplabv network miou miou ccf bdci dataset higher tradit deeplabv network miou outperform tradit deeplabv u net psp net macu net network | D007091;D058998 | T066 |
36,236,573 | intellig sensor base decis support system diagnos pulmonari ailment standard chest x ray scan academ health communiti pay much attent develop smart remot patient monitor sensor healthcar technolog analysi medic scan various studi integr sophist deep learn strategi smart monitor system need proactiv diagnost solut may employ epidemiolog scenario covid consequ work offer intellig medicar system iot empow deep learn base decis support system dss autom detect categor infecti diseas covid pneumothorax propos dss system evalu use three independ standard base chest x ray scan suggest dss predictor use identifi classifi area whole x ray scan abnorm thought attribut covid reach identif classif accuraci rate normal imag covid pneumothorax suggest dss system judgment depend individu chest x ray scan may made approxim result dss system describ studi forecast pace frame per second fps model near real time | D000086382;D011030 | T047;T067 |
36,236,571 | piezoelectr micromachin ultrason transduc integr helmholtz reson microlit size volum tunabl caviti studi piezoelectr micromachin ultrason transduc pmut integr microlit size volum tunabl helmholtz reson passiv helmholtz reson construct use su photolithographi defin squar open plate neck portion print hollow structur thread insert nut precis set screw form volum control caviti helmholtz reson fabric piezoelectr film act ultrason actuat attach surfac neck su plate experiment result show sound pressur level spl oper bandwidth could effect tune spl increas twofold bandwidth enhanc achiev set caviti length mm compar open caviti case modifi helmholtz reson model propos explain experiment result adjust factor effect mass viscous damper creat modifi exist paramet convent helmholtz reson model relationship adjust factor caviti length describ well use two term power seri curv modifi helmholtz reson model provid insight activ type helmholtz reson oper also provid use estim optim design fabric | D014159;D014465 | T074;T070 |
36,236,570 | smart wearabl detect occup physic fatigu literatur review today world chang dramat due influenc various factor whether due rapid develop technolog tool advanc telecommun method global econom social event reason almost everyth chang result concept job work chang well new work shift introduc offic longer place work done addit non stop activ societi increas stress pressur work caus fatigu spread worldwid becom global problem moreov medic proven persist fatigu caus serious diseas health problem therefor monitor detect fatigu workplac essenti improv worker safeti long term paper provid overview use smart wearabl devic monitor detect occup physic fatigu addit present discuss challeng hinder field highlight done advanc use smart wearabl workplac fatigu detect | D000076251;D017132 | T073;T082 |
36,236,569 | design implement ultrason flowmet base cross correl method ultrason flowmet play import role industri product aerospac field paper high precis ultrason flowmet base cross correl method design commerci finit element softwar comsol multiphys use simul propag process ultrason wave flow measur implement process cross correl algorithm simul python languag flowmet adopt cross correl algorithm improv measur accuraci ultrason time flight adopt method combin fpga embed microprocessor improv oper effici order verifi perform flowmet test flowmet nation institut metrolog self built test platform use still water drag method dynam volum method field comparison method respect result show flowmet abil test flow condit high flow veloc ms pipe diamet rang dndn absolut valu relat indic error exceed repeat exceed design ultrason flowmet high measur accuraci good repeat strong stabil wide applic rang | D045268 | T074 |
36,236,567 | indirect monitor frequenc multipl span bridg use data collect instrument train field case studi paper field studi carri monitor natur frequenc malahid viaduct bridg locat north dublin bridg includ seri simpli support span two collaps replac replac span stiffer other differ result higher natur frequenc indirect bridg monitor approach employ acceler respons instrument train use estim natur frequenc span viaduct show locat two replac span higher stiff indirect approach ensembl empir mode decomposit eemd base hilbert huang transform hht techniqu employ identifi natur frequenc span carri analys instantan frequenc if calcul intrins mode function averag if calcul use run instrument train vari carriag mass speed run use estim natur frequenc assess feasibl indirect approach bespok set direct measur taken use acceleromet attach success span viaduct free forc vibrat span use estim first natur frequenc frequenc obtain drive measur compar direct measur confirm effect indirect approach addit instantan amplitud drive signal use indic locat stiffer span final accuraci robust indirect approach monitor multi span bridg discuss | D000465;D014732 | T170;T070 |
36,236,565 | develop conductometr sensor base al ca dope zno detect formaldehyd present studi develop conductometr gas sensor base al ca dope zinc oxid composit final detect formaldehyd hcho low concentr air investig electr sens properti composit base zno dope differ load al andor ca evalu gas sens mechan al ca dope zinc oxid nanocomposit base sensor also discuss optim al ca zno sensor display formaldehyd respons ppm hchoair experiment low detect limit ppb hchoair oper temperatur c sensor also shown select hcho respect mani interfer indoor gase chang baselin resist opposit way compar target gas develop devic monitor hcho indoor workplac environ advantag simpl planar structur easili fabric mass product use low cost materi easi fabric method | D053761;D015034 | T121;T073;T197 |
36,236,564 | experiment analysi handcart push pull safeti industri environ use iot forc emg sensor relationship oper psycholog status pain syndrom non ergonom execut repetit physic task repres major caus work relat musculoskelet disord wmsd studi focus push pull p p industri handcart generic physic task present across mani industri aim investig depend p p execut oper psycholog status presenc pain syndrom upper limb spine develop acquisit system integr two three axi forc sensor place left right arm six electromyographi emg electrod place chest back hand flexor muscl conduct experi involv two group particip without increas psycholog score pain syndrom ten forc paramet left right side one emg paramet three differ muscl left right side two time domain paramet extract acquir signal data analysi show intergroup differ examin paramet especi forc integr valu emg mean absolut valu best knowledg first studi evalu composit effect pain syndrom spine mobil psycholog status particip execut p p task conclud signific impact p p task execut potenti risk wmsd futur work direct toward develop person risk assess system consid muscl group supplementari data deriv oper pose extract comput vision algorithm cognit paramet extract eeg sensor | D001132;D009140 | T047;T029 |
36,236,560 | comparison surrog behavior model power amplifi linear high spars data good approxim power amplifi pa behavior model requir precis baseband model mitig nonlinear sinc digit predistort dpd use provid pa linear framework necessari valid model figur merit support signal condit transmiss restrict field programm gate array fpga base testb develop measur wide band pa behavior use singl carrier quadratur amplitud modul qam multiplex orthogon frequenc divis multiplex ofdm base long term evolut lte stimulus differ bandwidth signal search provid heurist target approach model paper introduc featur extract concept find appropri complex solut consid high spars data issu amplitud amplitud amplitud phase pm model extract whose penalti associ overfit hardwar complex result function thus experiment result highlight model perform high spars data regim compar regress tree rt random forest rf cubic spline cs model accuraci capabl signal condit show reliabl valid low complex accord peak averag power ratio papr complementari cumul distribut function ccdf coeffici extract normal mean squar error nmse execut time figur merit present model provid comparison origin data aid compar dimens robust surrog model machin learn ml base ii cs interpol base high spars data present nmse cs interpol base also compar demonstr efficaci predict method lower converg time complex | D000669 | T073 |
36,236,558 | preliminari contribut toward risk base model flood manag plan use bim case studi lisbon prepar citi impact global warm becom major import adopt climat proof polici strategi respons climat chang becom fundament element citi plan end research consid multidisciplinari approach local scale abl connect urban plan architectur vital base consid coastal citi abil control consequ climat chang specif flood far scarciti research connect sea ground land survey studi could becom foundat refer coastlin settlement manag use bim found bim build inform model possibl tool manag coastal risk sinc combin crowdsourc data geometr inform model citi propos bim model includ topographi use themat map riverb model waterway model model aim facilit coordin across separ actor interest sinc urban area model alway updat improv focus case studi lisbon develop risk base map area close shorelin tagus river | D002958;D055868 | T064;T070 |
36,236,557 | acceler quantit ute cone imag use compress sens studi feasibl acceler quantit ultrashort echo time cone qute cone imag compress sens cs reconstruct investig qute cone sequenc variabl flip angl base ute map ute adiabat map ute quantit magnet transfer model macromolecular fraction mmf implement clinic mr system twenti healthi volunt recruit underw whole knee mri use qute cone sequenc k space data retrospect undersampl differ undersampl rate undersampl qute cone data reconstruct use zero fill cs reconstruct use cs reconstruct ute imag various paramet estim differ region interest roi tendon ligament menisci cartilag structur similar percentag error pearson correl calcul assess perform dramat reduc streak artifact improv ssim observ ute imag cs reconstruct mean ssim achiev cs reconstruct imag percentag error fulli sampl undersampl cs reconstruct imag undersampl ie acceler high linear correl observ gt qute paramet estim subject cs base reconstruct combin effici cone trajectori expect achiev clinic feasibl scan time qute imag | D021621;D008279 | T073;T059;T060;T170;T062 |
36,236,555 | compar analysi effici attent mechan traffic forecast without structur prior dot product attent power mechan captur contextu inform model build top acclaim state art perform various domain rang sequenc model visual task howev main bottleneck construct attent map quadrat respect number token sequenc consequ effici altern develop parallel recent perform compar contrast studi perform compar analysi effici attent mechan context pure attent base spatio tempor forecast model use traffic predict experi show method reduc train time infer time perform remain par baselin | D005544 | T090;T170;T079 |
36,236,554 | toward lightweight neural network garbag object detect recent year garbag classif becom hot topic china legisl garbag classif propos proper garbag classif improv recycl rate garbag protect environ save resourc order effect achiev garbag classif lightweight garbag object detect model base deep learn techniqu design develop studi locat classifi garbag object real time use embed devic focus problem low accuraci poor real time perform garbag classif propos lightweight garbag object detect model yolog yolo garbag detect base accur local recept field dilat run embed devic high speed high perform yolog improv yolov three key way includ design dcspresnet accur local recept field expans base dilat deform convolut network structur simplif use new activ function collect domest garbag imag dataset train test model final order compar perform differ yolog exist state art algorithm conduct comparison experi use uniform data set train model experiment result show yolog achiev ap comput gflop thus outperform yolov yolov yolov tini yolov term comprehens perform indic network propos paper detect domest garbag accur rapid provid foundat futur academ research engin applic | D000465;D016571 | T170 |
36,236,553 | yolo algorithm long term track detect escherichia coli differ depth microchannel base microspher posit assist effect evalu antibiot suscept test base bacteri solut great signific clinic diagnosi prevent antibiot abus appli microfluid chip detect platform detect method use microscop imag observ bacteria antibiot great speed detect time suitabl high throughput detect howev due influenc depth microchannel multipl layer bacteria focal depth microscop great affect count recognit accuraci increas difficulti reloc target bacteria well extract characterist bacteri liquid chang action antibiot focal depth target bacteria determin although z axi control help three dimension micro oper equip difficult oper long term chang target bacteria track quick accur paper yolov algorithm adopt accur identifi bacteria differ focus state multi layer bacteria z axi focal depth meantim certain amount microspher mix bacteria assist locat bacteria conveni track growth state bacteria long period recognit rate bacteria microspher high recognit accuraci count accuraci bacteria two recognit rate microspher respect much higher count accuraci bacteria microspher method enhanc depth field edf method moreov long term bacteri track detect target bacteria multipl z axi focal depth posit record aid microspher posit aid reconstruct focal depth posit reposit within h structur similar ssim microscop imag structur differ focal depth fluctuat differ time root mean squar error rmse fluctuat indic method also good reloc accuraci thus method provid basi rapid high throughput long term analysi microscop chang eg morpholog size bacteria detect addit antibiot differ concentr base microfluid channel futur | D004926;D004927 | T007;T047 |
36,236,550 | low cost low power water veloc sensor util acoust doppler measur current commerci sensor monitor water flow veloc expens bulki requir signific effort instal low cost sensor open possibl monitor storm wast water system much greater spatial tempor resolut without prohibit cost resourc invest aid work develop low cost low power veloc sensor base acoust doppler velocimetri sensor cost less usd open sourc open hardwar compact easili interfac wide rang data log system freeli avail sensor design price point current exist novelti enabl high resolut real time monitor scheme design capabl measur water veloc mms sensor characteris verifi field long term test final data test use evalu perform sensor real world scenario analysi conclud sensor capabl effect measur water veloc | D000162;D062065 | T069;T083;T070 |
36,236,549 | piezoelectr energi harvest rotat motion power industri mainten sensor industri forecast machineri failur could save signific time money mainten break predict aim work develop energi harvest system could theori power condit monitor sensor heavi machineri studi piezoelectr cantilev type energi harvest attach motor spun around differ rotat speed mass place tip cantilev mount point inward toward center axi motor point cantilev tip inward increas distanc center axi motor decreas natur reson frequenc signific thus enabl higher harvest energi level lower rotat frequenc motion cantilev also control alter movement space tip mass creat anoth possibl control cantilev dynam prevent overstress piezoelectr materi restrict movement tip mass also use harvest energi wider frequenc rang prevent harvest get trap stagnant posit highest calcul raw power w hz rotat frequenc measur cantilev outer dimens mm mm result suggest energi harvest system multipl cantilev could design replac batteri condit sensor monitor revolv machineri | D011211;D014159 | T074;T073 |
36,236,548 | novel non invas approach evalu soil moistur without soil disturb contactless ultrason system soil moistur consid key variabl govern terrestri ecosystem howev challeng preserv indigen soil characterist use convent soil moistur monitor method requir maximum soil contact overcom issu develop non destruct method evalu soil moistur use contactless ultrason system system design measur leaki rayleigh wave air soil joint half space influenc soil moistur leaki rayleigh wave explor sand silt clay control experiment design result show strong relationship energi amplitud leaki rayleigh wave soil moistur three soil case result explain reduc soil strength evapor process coars soil particl oppos fine soil particl evalu soil moistur base dynam paramet wave properti obtain observ leaki rayleigh wave use random forest model accuraci predict soil moistur except test data set soil type r rmse studi demonstr leaki rayleigh wave great potenti continu assess soil moistur variat without soil disturb | D017753;D012987 | T082;T167;T070 |
36,236,546 | lidson v lidar deep learn base green assist edg devic enhanc mobil visual impair billion peopl around world disabl among million visual impair blind number great increas due age chronic diseas poor environ health despit mani propos current devic system lack matur complet fulfil user requir satisfact increas research activ field requir order encourag develop commerci widespread accept low cost afford assist technolog visual impair disabl paper propos novel approach use lidar servo motor ultrason sensor collect data predict object use deep learn environ percept navig adopt approach use pair smart glass call lidson v enabl identif obstacl visual impair lidson system consist arduino uno edg comput devic integr smart glass smartphon app transmit data via bluetooth arduino gather data oper sensor smart glass detect obstacl use simpl data process provid buzzer feedback visual impair user smartphon applic collect data arduino detect classifi item spatial environ give spoken feedback user detect object comparison imag process base glass lidson use far less process time energi classifi obstacl use simpl lidar data accord sever integ measur comprehens describ propos system hardwar softwar design construct prototyp implement test real world environ use open platform weka tensorflow entir lidson system built afford shelf sensor microcontrol board cost less usd essenti provid design inexpens miniatur green devic built mount pair glass even wheelchair help visual impair approach enabl faster infer decis make use relat low energi smaller data size well faster communic edg fog cloud comput | D000077321;D006233;D012656;D019987;D014910 | T101;T074;T066 |
36,236,545 | vibrat base approach measur rail stress model first field test paper describ non invas inspect techniqu estim longitudin stress continu weld rail cwr infer rail neutral temperatur rnt ie temperatur net longitudin forc rail zero techniqu base use finit element method fem vibrat measur machin learn ml fem use model relationship boundari condit longitudin stress given cwr vibrat characterist mode shape frequenc rail result numer analysi use train ml algorithm test use field data obtain array acceleromet poll track interest studi present articl propos techniqu proven field experiment campaign conduct colorado commerci fem softwar use model rail track short rail segment repeat indefinit vari boundari condit stress three dataset prepar fed ml model develop use hyperparamet search optim techniqu k fold cross valid infer stress rnt frequenc vibrat extract time waveform obtain two acceleromet temporarili attach rail result experi demonstr success techniqu depend accuraci model abil proper identifi modeshap result also prove ml also abl predict success neutral temperatur test rail use limit number experiment data train | D000465;D014732 | T170;T070 |
36,236,544 | parallel classif model marin mammal sound base multi dimension featur extract data augment due poor visibl deep sea environ acoust signal often collect analyz explor behavior marin speci progress underwat signal acquisit technolog amount acoust data obtain ocean exceed limit human process manual design effici marin mammal classif algorithm becom research hotspot paper design classif model base multi channel parallel structur process multi dimension acoust featur extract audio sampl fuse predict result differ channel trainabl full connect layer use transfer learn obtain faster converg speed introduc data augment improv classif accuraci k fold cross valid method use segment data set comprehens evalu predict accuraci robust model evalu result show model achiev mean accuraci maintain standard deviat excel consist perform multipl test | D000465;D016571 | T170 |
36,236,542 | novel segment scheme multi probabl threshold human activ recognit use wearabl sensor recent year much research conduct time seri base human activ recognit har use wearabl sensor exist work har base manual label howev complet time serial signal contain differ type activ also includ mani transit atyp one thus effect filter activ becom signific problem paper novel machin learn base segment scheme multi probabl threshold propos har threshold segment ts slope area sa approach employ accord characterist small fluctuat static activ signal typic peak trough period like one addit multi label weight probabl mlwp model propos estim probabl activ har error signific decreas propos model solv problem fix window usual contain multipl kind activ unknown activ accur reject reduc impact compar exist scheme comput simul reveal propos model maintain high perform use uci pamap dataset averag har accuraci abl reach respect | D006802;D000076251 | T073;T052 |
36,236,541 | threshold base user assist cooper relay beamspac massiv mimo noma system incorpor user assist cooper relay beamspac massiv multipl input multipl output mmimo non orthogon multipl access noma system extend coverag area improv spectral energi effici millimet wave mmwave communic dynam cluster mobil user termin mut form within beam propos threshold base user assist cooper relay beamspac mmimo noma system downlink scenario specif intermedi mut next generat base station gnb cell edg mut becom relay mut success decod signal cell edg mut meet predetermin signal interfer plus nois ratio sinr threshold zero forc zf precod iter power alloc use minim inter intra beam interfer maxim system sum rate evalu perform system delay intoler cell edg mut scenario moreov outag probabl cell edg mut propos scheme investig analyt express deriv simul result confirm propos threshold base user assist cooper relay beamspac mmimo noma system outperform user assist cooper relay beamspac mmimo noma beamspac mimo noma beamspac mimo orthogon multipl access oma system term spectrum effici energi effici outag probabl | D009625 | T047 |
36,236,539 | novel planar strain sensor design captur dimension fingertip forc patient affect hand paralysi assess therapi individu hand paresi requir forc sens approach measur wide rang finger forc multipl dimens present novel strain gaug forc sensor degre freedom dof design use hand assess rehabilit devic sensor featur fiberglass print circuit board substrat eight strain gaug bond circuiti sensor rout direct board secur larger rehabilit devic via aluminum frame design sens packag character weight capac resolut requir furthermor test sensor calibr three axi configur valid larger spheric workspac understand accur precis sensor sensor slight shortcom valid error satisfi precis calibr accuraci fine sens requir orthogon load structur specif met sensor therefor great candid sens technolog rehabilit devic assess dexter patient impair hand function | D000535;D006225 | T121;T023;T196 |
36,236,538 | detect pest infest forest damag multispectr satellit imageri improv unet plant pest primari biolog threat agricultur forestri product well forest ecosystem monitor forest pest damag via satellit imag crucial develop prevent control strategi previous studi util deep learn monitor pest infest damag satellit imageri adopt rgb imag multispectr imageri veget indic use multispectr imag veget indic contain wealth use inform detect plant health improv precis pest damag detect aim studi improv forest pest infest area segment combin multispectr veget indic rgb inform deep learn also propos new imag segment method base unet attent mechan modul detect forest damag induc bark beetl aspen leaf miner sentinel imag resnest use featur extract backbon attent mechan scse modul introduc decod phase improv imag segment result use sentinel imageri produc dataset base forest health damag data gather ministri forest land natur resourc oper rural develop flnrord british columbia bc canada aerial overview survey ao dataset contain origin sentinel band veget indic experiment result confirm signific veget indic multispectr data enhanc segment effect result demonstr propos method exhibit better segment qualiti accur quantit indic overal accuraci comparison state art pest area segment method | D017753;D063809 | T082;T068;T073;T070 |
36,236,536 | anomali detect algorithm base ensembl learn g environ advent digit inform age new data servic virtual realiti industri internet cloud comput prolifer recent year result increas oper demand g bearer network provid featur high transmiss capac ultra long transmiss distanc network slice intellig manag control softwar defin network new network architectur intend increas network flexibl agil better satisfi demand g network network slice nevertheless softwar defin network still face challeng network intrus propos abnorm traffic detect method base stack method self attent mechan make shortcom inabl track long term depend data sampl ensembl learn method util self attent mechan convolut network automat learn long term associ traffic sampl provid downstream task sampl embed addit design novel stack ensembl method comput sampl embed predict valu heterogen base learner fusion modul obtain final outlier result paper conduct experi abnorm traffic dataset softwar defin network environ calcul precis recal f score compar analyz algorithm experiment result show method design paper achiev multipl indic precis recal f score respect better comparison method | D000465;D012984 | T090;T170;T073 |
36,236,535 | softmatch compar scanpath use combinatori spatio tempor sequenc fractal curv recent studi match eye gaze pattern other contain research heavili reliant string edit method borrow earli work bioinformat previous studi shown string edit method suscept fals negat result match mutat gene unord region interest scanpath even new method emerg match amino acid use novel combinatori techniqu scanpath match still limit tradit collinear approach approach reduc abil discrimin free view scanpath two peopl look stimulus due heavi weight place linear overcom limit introduc new method call softmatch compar pair scanpath softmatch diverg tradit scanpath match two differ way first preserv local use fractal curv reduc dimension cartesian x coordin h hilbert distanc second take combinatori approach fixat match use discret frchet distanc measur segment scanpath fixat sequenc match sequenc fixat time loos acronym softmatch result indic high degre statist substant signific score match scanpath made free form view unfamiliar stimuli applic method use better understand bottom perceptu process extend scanpath outlier detect expertis analysi patholog screen salienc predict | D005403;D017709 | T042;T077 |
36,236,534 | coupl visual inerti measur unit method locat map coal mine tunnel mobil robot move fast scene poor light condit often caus loss visual featur track coal mine tunnel ground often bumpi light uneven movement mobil robot scene violent bump local technolog visual featur great affect illumin speed camera movement solv local map problem environ similar underground coal mine tunnel improv local map algorithm base monocular camera inerti measur unit imu featur match method combin point line featur design improv robust algorithm presenc degrad scene structur insuffici illumin tight coupl method use establish visual featur constraint imu pre integr constraint keyfram nonlinear optim algorithm base slide window use accomplish state estim extens simul practic environ verif show improv simultan local map slam system monocular camera imu fusion achiev accur autonom local map construct scene insuffici light coal mine tunnel | D000465;D003031 | T121;T170;T109;T167 |
36,236,533 | machin learn base peripher arteri diseas identif use laboratori base gait data peripher arteri diseas pad manifest atherosclerosi limit blood flow leg caus chang muscl structur function gait perform pad underdiagnos delay treatment worsen clinic outcom overcom challeng purpos studi develop machin learn ml model distinguish individu without pad first step use ml identifi pad risk earli built ml model base previous acquir overground walk biomechan data patient pad healthi control gait signatur character use ankl knee hip joint angl torqu power well ground reaction forc grf ml abl classifi without pad use neural network random forest algorithm accuraci matthew correl coeffici use laboratori base gait variabl moreov model use grf variabl provid accuraci matthew correl coeffici result indic ml model classifi without pad use gait signatur accept perform result also show ml gait signatur model use grf featur deliv inform data pad classif | D005684;D058729 | T047;T033 |
36,236,532 | applic continu wavelet transform artifici naural network automat radar signal recognit articl aim propos algorithm automat recognit select radar signal algorithm find applic area electron warfar ew automat recognit type intra puls modul type emitt oper mode aid decis make process simul carri includ analysi classif possibl linear frequenc modul puls waveform lfmpw step frequenc modul puls waveform sfmpw phase code puls waveform pcpw rectangular puls waveform rpw frequenc modul continu wave fmcw continu wave cw step frequenc continu wave sfcw phase code continu waveform pccw algorithm propos paper base use continu wavelet transform cwt coeffici higher order statist hos featur determin select signal princip compon analysi pca method use dimension reduct artifici neural network use classifi simul studi took account presenc nois interfer signal nois ratio snr rang db final obtain classif effici present form confus matrix simul result show high recognit test accuraci signal nois ratio greater db articl also deal select type paramet wavelet author also point problem encount research exampl solv | D011825;D058067 | T062;T073 |
36,236,531 | internet thing secur solut survey overwhelm accept grow need internet thing iot product aspect everyday live creat promis prospect involv human data procedur vast area creat opportun home industri make autom lifecycl human life involv enorm applic intellig transport intellig healthcar smart grid smart citi etc thrive surfac creat affect societi economi environ polit health divers secur threat general iot devic suscept secur breach develop industri system could pose devast secur vulner build reliabl secur shield challeng encount must embrac therefor survey paper primarili aim assist research classifi attacksvulner base object method attack relev countermeasur provid kind attack work case studi import applic iot highlight concern secur solut survey secur solut limit tradit secret key base cryptograph solut moreov physic unclon function puf base solut blockchain illustr pros con secur solut also discuss furthermor challeng recommend present work | D000081186;D000080487 | T170;T066 |
36,236,530 | review human comfort factor measur improv human robot collabor develop robot technolog collabor robot cobot applic human robot collabor hrc grow past decad despit tremend effort academia industri overal usag accept cobot still high expect one major affect factor comfort human hrc usual less emphas cobot develop howev critic user accept hrc therefor paper give review human comfort hrc includ influenti factor human comfort measur human comfort term subject object manner human comfort improv approach context hrc discuss topic also conduct base review analysi | D012371 | T090 |
36,236,528 | pedestrian origin destin estim base multi camera person identif pedestrian origin destin estim record traffic flow origin destin essenti manag pedestrian facil includ pedestrian flow simul plan phase crowd control oper phase howev current data collect techniqu survey mobil sens use gps wi fi bluetooth smart card data disadvantag either time consum cost provid complet inform pedestrian facil without entranc exit pedestrian flow insid facil due full coverag cctv camera huge potenti imag process techniqu address challeng pedestrian estim propos imag base estim framework identifi person disjoint camera view trajectori ident accur generat state art deep neural network dnns person id differ congest level compar improv final matrix base trajectori generat resid time calcul provid recommend pedestrian facil improv factor affect accuraci framework discuss paper believ could provid new insight stimul research applic internet camera intellig transport infrastructur manag | D000069636 | T098 |
36,236,527 | innov fusion base scenario improv land crop map accuraci accuraci land crop map obtain satellit imag depend type featur select algorithm classifi algorithm differ effici differ condit therefor develop suitabl strategi combin capabl differ algorithm prepar land crop map higher accuraci use object studi develop fusion base framework improv land crop map accuraci first featur retriev use sentinel sentinel landsat imageri train data various featur select algorithm includ recurs featur elimin rfe random forest rf boruta use optim featur select various classifi includ artifici neural network ann support vector machin svm rf implement creat map land crop reli optim featur train data order increas result accuraci map land crop deriv sever scenario fuse use fusion base vote strategi level decis new map land crop classif uncertainti map prepar subsequ perform differ scenario evalu compar among featur select algorithm rf accuraci higher rfe boruta moreov effici rf higher svm ann overal accuraci vote scenario higher scenario find research demonstr combin featur capabl extract sensor differ spectral rang differ featur select algorithm classifi improv land crop classif accuraci | D000465;D060388 | T170;T066;T081 |
36,236,525 | character infant general movement use commerci rgb depth sensor deep neural network track process tool exploratori studi cerebr palsi common childhood neuromotor disord often diagnos visual assess general movement gm infanc skill requir extens train thus difficult implement larg scale autom analysi gm perform use low cost instrument home may use estim quantit metric predict movement disord studi explor infant gm may success evalu familiar environ process trajectori point interest poi obtain record singl commerci rgb sensor rgb video process use open sourc markerless motion track method allow estim trajectori select poi purpos develop method allow reconstruct trajectori make use data record depth sensor eight infant gm record home month age eight gm metric propos literatur addit novel metric estim poi trajectori timepoint pediatr neurologist physiatrist provid overal clinic evalu infant video subsequ comparison metric clinic evalu perform result demonstr gm metric may meaning estim potenti use earli identif movement disord | D002547;D009069 | T047;T019 |
36,236,524 | droplet deposit distribut predict method six rotor plant protect uav base invers distanc weight aim work establish method real time calcul droplet deposit distribut six rotor plant protect unman aerial vehicl uav numer simul airflow field carri use comput fluid dynam cfd airflow field distribut obtain seven flight speed six flight height seven crosswind speed relat error verifi accuraci numer model within spatial point wind speed test simul valu numer simul result show improv uav flight speed crosswind relat airflow produc vortex downwash wind field uav reduc stabil downwash wind field discret droplet phase introduc flow field ground region divid use small grid statist calcul droplet deposit rate within grid yield distribut droplet differ paramet combin statist result show relat airflow crosswind caus droplet convolut droplet drift increas actual oper uav flight speed well control condit low environment wind reduc droplet drift rate improv util rate pesticid base distribut differ paramet combin one droplet deposit predict method establish use invers distanc weight idw propos method lay foundat cumul calcul droplet deposit distribut continu oper plant protect uav provid basi object evalu oper qualiti plant protect uav optim set oper paramet | D010575;D014919 | T131;T070 |
36,236,523 | wsneap effici authent protocol iiot orient wireless sensor network develop industri internet thing iiot industri wireless sensor need upload collect privat data cloud server result larg amount privat data expos internet privat data vulner hack mani complex wireless sensor authent protocol propos paper propos effici authent protocol iiot orient wireless sensor network protocol introduc puf chip use bloom filter save queri challeng respons pair generat puf chip ensur secur physic layer devic reduc comput cost communic cost wireless sensor side protocol introduc pre authent mechan achiev continu authent gateway cloud server overal comput cost protocol reduc formal secur analysi inform secur analysi prove propos protocol secur featur implement various secur primit use miracl cryptograph librari gmp larg number librari propos protocol compar depth relat work detail experi show propos protocol signific reduc comput cost communic cost wireless sensor side overal comput cost protocol | D016494;D000080487 | T078;T170;T057;T097;T080 |
36,236,522 | robust lidar inerti odometri ground condit percept optim algorithm ugv unman ground vehicl ugv make progress mani applic scenario recent year explor unknown wild terrain work precis agricultur serv emerg rescu due complex ground condit changeabl surround unstructur environ challeng ugv obtain robust accur state estim use sensor fusion odometri without prior percept optim specif scenario paper base error state kalman filter eskf fusion model propos robust lidar inerti odometri novel ground condit percept optim algorithm specif design ugv probabl distribut gain raw inerti measur unit imu measur certain time period state estim eskf util evalu flat ground condit real time analyz relationship current ground condit accuraci state estim tight coupl lidar inerti odometri dynam optim adjust relat paramet process algorithm lidar point obtain robust accur ego motion state estim ugv method valid various type environ changeabl ground condit robust accuraci shown consist accur state estim differ ground condit compar state art lidar inerti odometri system | D000465;D010465 | T170;T041 |
36,236,521 | progress biosensor point care diagnosi covid coronavirus diseas covid high virul infect caus pandem sinc earli diagnosi diseas recogn one import approach minim patholog impact spread infect point care test prove substanti analyt tool especi later flow immunoassay later flow test serv purpos last year biosensor gain popular simpl high sensit accur analyt devic compos select molecul antibodi antigen sensor platform biosensor would advanc altern current point care test covid diagnosi standard laboratori method well recent discoveri relat point care diagnost test covid develop biosensor specif antibodi specif virus part genet inform review | D015374;D000086382 | T059;T047;T075;T067 |
36,236,520 | design analysi quartz crystal microbal new ring shape interdigit electrod paper new type ring shape interdigit electrod propos improv accuraci repeat quartz crystal microbal influenc differ type singl finger dot finger dot doubl finger electrod mass sensit distribut well optim proport finger gap width obtain multi physic coupl simul result show design criteria interdigit electrod chang increas number finger gap width obey decreas order central edg twice width finger width outermost finger radius middl dot electrod maintain time total electrod radius experi carri verifi quartz wafer dot doubl finger electrod high qualiti factor less modal coupl satisfi engin applic well conclus studi provid design idea electrod maintain uniform distribut quartz crystal microbal mass sensit | D015374;D058525 | T073;T059;T063;T075;T062 |
36,236,519 | verif system ship posit keep equip set anchor unityd modern comput specialis softwar abl simul ocean wave sea current action wind graviti ship vehicl high level program languag use type softwar read inform navig devic connect comput eg via serial port proceed use raw data control algorithm desktop softwar simul use data addit electron devic pressur sensor temperatur sensor etc thus possibl conduct real time communic plc programm logic control use simul articl user interfac design unityd present user interfac abl read data navig devic use ship posit control algorithm verif algorithm occur research real ship use anchor base posit system use data obtain real ship mathemat model anchor winch develop next mathemat model implement simul develop unityd simul contain environment condit research real ship mathemat model anchor winch implement develop simul allow futur research anchor base posit system eg differ environment condit research result shift ship posit degre differ arriv time target point real ship virtual ship differ posit deviat | D008962;D012767 | T093;T062;T170;T073 |
36,236,518 | ship fire detect base improv yolo algorithm lightweight convolut neural network model ship fire one greatest danger ship navig safeti nevertheless typic detect method limit detect effect accuraci due distanc restrict ship motion although issu address imag recognit algorithm base deep learn comput complex effici ship detect tough paper propos lightweight target identif techniqu base modifi yolov tini algorithm precis effici detect ship fire take account distinct characterist ship fire marin environ initi multi scale detect techniqu appli broaden detect rang integr deep semant inform therebi enhanc featur inform small target obscur object improv detect precis propos algorithm employ se attent mechan inter channel featur fusion improv capabl featur extract precis ship fire detect last least pictur transform migrat learn ad small ship fire dataset acceler converg pace improv converg effect reduc dataset depend simul experi reveal propos yolov tini se model outperform benchmark algorithm term ship fire detect accuraci detect effici satisfi real time ship fire warn criteria demand maritim environ | D005390;D012767 | T093;T073;T067 |
36,236,517 | hint base imag color base hierarch vision transform hint base imag color imag imag translat task aim creat full color imag input lumin imag small set color valu pixel given hint though tradit deep learn base method propos literatur base convolut neural network cnns strong spatial local due convolut oper often caus non trivial visual artifact color result fals color color bleed artifact overcom limit studi propos vision transform base color network propos hint base color network hierarch vision transform architectur form encod decod structur base transform block propos method use transform block learn rich long rang depend achiev visual plausibl color result even small number color hint verif experi result reveal propos transform model outperform convent cnn base model addit qualit analyz effect long rang depend transform model hint base imag color | D000465;D016571 | T170 |
36,236,516 | artifici intellig base detect human embryo compon assist reproduct vitro fertil assist reproduct technolog help human address infertil use differ medic procedur help success pregnanc vitro fertil ivf one assist reproduct method sperm egg combin outsid human bodi special environ kept growth assist reproduct technolog help human address infertil use differ medic procedur help success pregnanc morpholog embryolog compon high relat success assist reproduct procedur approxim day embryo transform blastocyst prevent multipl birth risk increas chanc pregnanc embryologist manual analyz blastocyst compon select valuabl embryo transfer women uterus manual microscop analysi blastocyst compon trophectoderm zona pellucida blastocoel inner cell mass time consum requir keen expertis select viabl embryo artifici intellig eas medic procedur success implement deep learn algorithm mimic medic doctor knowledg provid better diagnost procedur help reduc diagnost burden deep learn base automat detect blastocyst compon help analyz morpholog properti select viabl embryo research present deep learn base embryo compon segment network ec net accur detect trophectoderm zona pellucida blastocoel inner cell mass embryolog analysi propos method ec net base shallow deep segment network use two separ stream produc base convolut block depth wise separ convolut block stream dens concaten combin two dens skip path produc power featur upsampl propos ec net evalu public avail microscop blastocyst imag dataset experiment segment result confirm efficaci propos method propos ec net provid mean jaccard index mean ji embryolog analysi | D001185;D007246 | T046;T073;T066;T090;T033 |
36,236,515 | data fusion base musculoskelet synergi grasp hand hypothesi central nervous system cns make use synergi movement primit achiev simpl complex movement inspir investig differ type synergi kinemat muscl synergi extens studi literatur studi compar combin type synergi control coordin human hand paper synergi extract first independ call kinemat muscl synergi combin data fusion call musculoskelet synergi activ daili live individu use princip compon analysi pca independ compon analysi ica weight linear combin musculoskelet synergi record kinemat record muscl activ reconstruct perform musculoskelet synergi reconstruct movement compar synergi report previous literatur us other result indic musculoskelet synergi perform better synergi extract without fusion attribut improv perform musculoskelet synergi generat basi cross inform muscl kinemat activ moreov synergi extract use ica perform better synergi extract use pca musculoskelet synergi possibl improv capabl current methodolog use control high dimension prosthet exoskeleton | D000203;D018737 | T056;T081;T040;T080 |
36,236,513 | integr framework multi state driver monitor use heterogen loss attent base featur decoupl multi state driver monitor key techniqu build human centric intellig drive system paper present integr visual base multi state driver monitor framework incorpor head rotat gaze blink yawn solv challeng head pose gaze estim paper propos unifi network architectur tackl estim soft classif task featur decoupl modul develop decoupl extract featur differ axi domain furthermor cascad cross entropi design restrict larg deviat train phase combin featur form heterogen loss function addit gaze consist use optim estim also inform model architectur design gaze estim task final propos method verifi sever wide use benchmark dataset comprehens experi conduct evalu propos method experiment result show propos method could achiev state art perform compar method | D001334 | T056 |
36,236,512 | research wind power short term forecast method base tempor convolut neural network variat modal decomposit wind energi reserv larg worldwid random volatil hinder wind power develop promot util wind energi improv accuraci wind power predict comprehens consid influenc wind farm environment factor histor power wind power generat paper present short term wind power predict model base time convolut neural network tcn variat mode decomposit vmd first due non smooth characterist wind farm environment data paper use vmd decompos data environment variabl reduc influenc random nois data predict model modal compon rich featur inform extract accord pearson correl coeffici maxim inform coeffici mic modal compon power third predict model base tcn train accord prefer modal compon histor power data achiev accur short term wind power predict paper model train test public wind power dataset provid spanish power compani simul result show model higher predict accuraci mape r respect compar convent long short term neural network lstm model model paper good predict accuraci robust | D000465;D016571 | T170 |
36,236,511 | associ self report prior night sleep singl task gait healthi young adult studi use machin learn failur obtain recommend h sleep associ injuri youth adult howev research influenc prior night sleep gait conduct older adult clinic popul therefor object studi identifi individu experi partial sleep depriv andor sleep extens prior night use singl task gait particip n age year femal agre particip studi self report sleep durat night prior test collect gait data collect inerti sensor min walk test group differ lt h gt h poor sleeper h good sleeper gait characterist assess use machin learn post hoc ancova result indic correl r gait paramet prior night sleep accur machin learn model random forest classifi use top featur mean accuraci find suggest good sleeper asymmetr gait pattern better maintain gait speed poor sleeper research larger subject size need develop accur machin learn model identifi prior night sleep use singl task gait | D012890;D012892 | T047;T046;T040;T033 |
36,236,510 | earabl omega novel clench interfac use ear canal sens human metacarpophalang joint control function electr stimul background mouth free interfac requir function electr stimul fes peopl spinal cord injuri develop novel system clench human metacarpophalang mp joint use earphon type ear canal movement sensor experi control joint angl joint stiff perform use develop system method propos fes use equilibrium point control signal stiff control signal electr agonist antagonist ratio electr agonist antagonist sum angl sensor use acquir joint angl system identif util measur joint stiff use extern forc robot arm experi includ six five subject respect result joint angl could control well clench hysteresi delay three subject could control relat well hyperextens subject reveal calibr problem chang characterist human mp joint caus hyperextens joint stiff increas clench amplitud five subject addit result indic viscos control conclus develop system control joint angl stiff futur research develop method show system control equilibrium point stiff simultan | D004424;D013119 | T023;T037 |
36,236,509 | estim region methan emiss factor energi agricultur sector sourc use portabl measur system case studi denver julesburg basin methan ch power greenhous gas ghg identifi key target emiss reduct pari agreement current clear effort focus make greatest impact current activ data standard emiss factor ef use generat ghg emiss inventori mani ef global uniform account region variabl industri agricultur practic andor regul region ef deriv top emiss measur use make bespok region ghg emiss inventori account geopolit social variabl howev larg scale top approach campaign requir signific invest address lower cost drive survey ds identifi viabl altern establish method dss take top measur mani emiss sourc relat short period time albeit higher uncertainti investig use portabl measur system km ds conduct throughout denver julesburg basin djb djb cover area km north denver co dens popul ch emiss sourc includ oil gas g oper agricultur oper ago lake reservoir ds individu ch emiss sourc detect sourc ago g oper natur sourc respect methan emiss sourc quantifi use downwind concentr meteorolog data ago g oper repres near ch emiss djb account total emiss respect oper similar emiss sourc group togeth averag facil emiss estim generat agricultur sourc emiss feedlot cattl dairi cow sheep estim g ch head h agre publish valu taken focus measur campaign similar g averag emiss well pad compressor station gas process plant kg ch facil h reason agreement emiss estim intens measur campaign comparison basin wide g emiss measur taken decad ago show decreas factor three feasibl explain chang g regul past year emiss ago remain constant time period data suggest dss could low cost altern tradit measur campaign use screen mani emiss sourc within region deriv repres region specif time sensit ef key benefit ds mani region screen emiss reduct target identifi region ef notic larger region nation global averag | D000393;D000074382 | T131;T197 |
36,236,508 | review mem vibrat gyroscop reliabl issu harsh environ micro electromechan system mem vibrat gyroscop gain lot attent last two decad low power consumpt easi integr low fabric cost usag gyroscop equip inerti measur unit increas tremend applic rang household devic smart electron militari equip howev reliabl issu still concern oper inerti sensor harsh environ control movement align mini satellit space track firefight elev temperatur assist aircraft navig gusti turbul air review paper focus key fundament mem vibrat gyroscop first discuss popular design includ tune fork gimbal vibrat ring multi axi gyroscop investig bias stabil angl random walk scale factor perform paramet affect harsh environ discuss reliabl issu gyroscop | D055617 | T068;T170 |
36,236,507 | salienc detect light field imag fuse focus degre grabcut light field imag salienc detect task redund cue introduc due comput method inevit lead inaccur boundari segment detect result problem chain block effect tackl issu propos method salient object detect sod light field imag fuse focus grabcut method improv light field focus calcul base spatial domain perform secondari blur process focus imag effect suppress focus inform focus area differ focus imag aim redund focus cue generat multipl foreground imag use optim singl foreground imag generat focus cue addit aim fusion various cue light field complex scene grabcut algorithm combin focus cue guid generat color cue realiz automat salienc target segment imag foreground extens experi conduct light field dataset demonstr algorithm effect segment salient target area background area light field imag outlin salient object clear compar tradit grabcut algorithm focus degre use instead artifici interact initi grabcut achiev automat salienc segment | D000465;D003463 | T078;T170 |
36,236,506 | boostedenml effici techniqu detect cyberattack iot system use boost ensembl machin learn follow recent advanc wireless communic lead increas internet thing iot system mani secur threat current ravag iot system caus harm inform consid vast applic area iot system ensur cyberattack holist detect avoid harm paramount machin learn ml algorithm demonstr high capac help mitig attack iot devic edg system reason accuraci howev dynam oper intrud iot network requir improv id model capabl detect multipl attack higher detect rate lower comput resourc requir one challeng iot system mani ensembl method use differ ml classifi includ decis tree random forest propos id model iot environ boost method one approach use design ensembl classifi paper propos effici method detect cyberattack network intrus base boost ml classifi propos model name boostedenml first train six differ ml classifi dt rf et lgbm ad xgb obtain ensembl use stack method anoth major vote approach two differ dataset contain high profil attack includ distribut denial servic ddos denial servic dos botnet infiltr web attack heartble portscan botnet use train evalu test id model ensur obtain holist effici model perform data balanc synthet minor oversampl techniqu smote adapt synthet adasyn techniqu use stratifi k fold split data train valid test set base best two model construct propos boostedensml model use lightgbm xgboost combin two classifi give lightweight yet effici model part target research experiment result show boostedensml outperform exist ensembl model term accuraci precis recal f score area curv auc reach case select dataset multiclass classif | D000080487 | T170 |
36,236,505 | thermal infrar track method base effici global inform percept solv insuffici abil current thermal infrar tir track method resist occlus interfer similar target propos tir track method base effici global inform percept order effici obtain global semant inform imag use transform structur featur extract fusion featur extract process focal transform structur use improv effici remot inform model high similar human attent mechan featur fusion process supplement relat posit encod standard transform structur allow model continu consid influenc posit relationship learn process also general captur differ posit inform differ input sequenc thus make transform structur model semant inform contain imag effici improv track accuraci robust heterogen bi predict head util object predict process fulli connect sub network respons classif predict foreground background convolut sub network respons regress predict object bound box order allevi contradict vast demand train data transform model insuffici scale tir track dataset lasot tir dataset generat generat adversari network network train method achiev best perform compar state art tracker vot tir vot tir ptb tir lsotb tir dataset perform outstand especi deal sever occlus interfer similar object | D010465;D012660 | T078;T041 |
36,236,504 | novel sensor system vivo percept reconstruct base long short term memori network monitor bodili pressur could provid valuabl medic inform doctor patient long term implant vivo sensor high desir situat percept reconstruct need particular fecal incontin artifici anal sphincter without percept could remind patient defec even caus ischem tissu necrosi due uncontrol clamp pressur address issu novel self packag strain gaug sensor system design vivo percept reconstruct addit long short term memori lstm network show excel perform process time seri relat featur fit properti use articl improv predict accuraci percept model propos system test compar tradit linear regress lr approach use data vitro experi result show root mean squar error rmse reduc demonstr predict accuraci propos lstm model higher lr model reach accur predict amount intestin content furthermor outcom vivo experi show robust novel sensor system base long short term memori network verifi experi limit data | D005242;D008570 | T047;T184;T041 |
36,236,503 | classif wheelchair relat shoulder load activ wearabl sensor data machin learn approach shoulder problem pain patholog high preval manual wheelchair user spinal cord injuri problem lead limit activ daili life adl labor leisur particip increas health care cost shoulder problem often associ long term relianc upper limb accompani shoulder load make estim daili shoulder load crucial know adl perform execut free live environ term magnitud frequenc durat aim studi develop valid methodolog classif wheelchair relat shoulder load adl sl adl wearabl sensor data ten abl bodi particip equip five shimmer sensor wheelchair upper extrem perform eight relev sl adl deep learn network use bidirect long short term memori network train sensor data acceler gyroscop signal emg use video annot activ target overal train algorithm perform well accuraci specif reduc input train network data one sensor overal perform decreas around perform measur use forearm sensor data led better perform use upper arm sensor data conclud generaliz algorithm could train deep learn network classifi wheelchair relat sl adl wearabl sensor data | D000076251;D014910 | T074;T073 |
36,236,502 | practic method blind pixel detect push broom thermal infrar hyperspectr imag thermal infrar hyperspectr imag one frontier payload current hyperspectr remot sens research broad applic prospect land ocean temperatur invers environment monitor field howev due influenc product process infrar focal plane array characterist materi infrar focal plane array inevit blind pixel result spectral distort data even invalid data limit applic thermal infrar hyperspectr data current blind pixel detect method base spatial dimens imag process singl band area imag push broom thermal infrar hyperspectr imag work complet differ convent area array thermal imag one row data obtain per scan therefor current method direct appli blind pixel detect push broom thermal infrar hyperspectr imag base imag principl push broom thermal infrar hyperspectr imag propos practic blind pixel detect method method consist two stage detect repair four common type blind pixel dead pixel dark current pixel blink pixel nois pixel first stage dead pixel dark current pixel low spectral respons rate detect spectral filter detect nois pixel detect spatial nois detect dark current pixel negat respons slope detect respons slope detect second stage accord random appear blink pixel spectral filter detect use detect repair spectral anomali caus blink pixel line line order verifi effect propos method flight test carri use airborn thermal infrar hyperspectr imag system athi latest thermal infrar imag china data acquisit result show method propos paper accur detect repair blind pixel thus effect elimin spectral anomali signific improv imag qualiti | D029865 | T002 |
36,236,501 | deep learn base remain use life estim bear time frequenc inform modern industri product predict abil remain use life bear direct affect safeti stabil system tradit method requir rigor physic model perform poor complex system paper end end remain use life predict method propos use short time fourier transform stft preprocess consid time correl signal sequenc long short term memori network design cnn incorpor convolut block attent modul understand decis make process network interpret level experi carri phm dataset compar method result prove effect method | D000077321 | T066 |
36,236,500 | toward polici develop methodolog human centr iot collect embed ethic concept smart internet connect devic make behav human centr manner ie ethic social accept manner receiv signific attent softwar industri make smart devic behav human center manner import develop methodolog defin smart devic key role map socio ethic administr polici paper propos polici develop methodolog make smart devic human centr follow four phase ie concept develop defin map polici implement process polici deploy devic suggest methodolog may use varieti situat smart devic interact peopl illustr propos methodolog appli three differ set includ supermarket children hospit earli learn center phase defin methodolog follow applic methodolog smart internet connect devic includ robot smart camera smart speaker shown signific result observ devic behav human centric way perform core function adher socio ethic polici | D057766;D011050 | T170;T057;T064 |
36,236,499 | survey collis avoid system underground mine sens protocol grow number unintent interact occur underground mine collis avoid system cas establish mainten becom urgent need mine industri enhanc risk profil improv construct safeti usual collis accid divid three differ categori line involv particip infrastructur condit accid pose great risk financi cost mine compani even caus casualti detail paper present intens studi survey posit techniqu includ rang algorithm accommod demand various proxim sensor improv capabl situat awar exploit import communic system preval low power wide area technolog relat communic protocol effect communic system decid facilit success final integr system use fundament address problem collis avoid purpos collabor communic system execut depart seri systemat comparison pertin technolog algorithm given near end follow brief discuss best choic among option propos solut overal end end delay minimis nanosecond localis accuraci achiev centimetr level oper rang | D000465;D008906 | T170;T057 |
36,236,497 | later flow immunoassay reader technolog quantit point care test due recent pandem caus coronavirus diseas covid later flow immunoassay use rapid antigen test popular ever howev histori later flow immunoassay year old origin purpos use covid rapid antigen test pregnanc test qualit detect target analyt recent demand quantit analysi later flow immunoassay increas various field later flow immunoassay quantit detect use various materi sensor technolog introduc reader analyz develop quantit analysi reader high anticip futur develop line technolog advanc optic magnet field phototherm electrochem sensor trend weight reduct miniatur cost reduct system addit sens process communic function portabl person devic smartphon use tool quantit analysi later flow immunoassay result later flow immunoassay effici achiev goal rapid diagnosi point care test reader use quantif later flow immunoassay classifi accord adopt sensor technolog research trend review paper develop quantit analysi system often carri assay aspect reader also assay develop case review necessari addit system quantit analysi covid recent gain import introduc separ section | D000086382 | T047;T067 |
36,236,496 | parallel cross convolut recurr neural network automat imbalanc ecg arrhythmia detect continu wavelet transform automat detect arrhythmia use electrocardiogram ecg deep learn dl import reduc global death rate cardiovascular diseas cvd previous studi automat arrhythmia detect reli larg various ecg featur achiev consider classif accuraci use dl base model howev previous research ignor multi class imbalanc problem ecg arrhythmia detect therefor remain challeng improv classif perform dl base model paper propos novel parallel cross convolut recurr neural network order improv arrhythmia detect perform imbalanc ecg signal propos model incorpor recurr neural network two dimension convolut neural network cnn effect learn tempor characterist rich spatial inform raw ecg signal continu wavelet transform cwt use transform ecg signal scalogram compos time frequenc compon subsequ cnn learn spatial inform scalogram propos model effici learn featur imbalanc sampl also signific improv model converg higher accuraci overal perform propos model evalu base mit bih arrhythmia dataset detail analysi evalu metric reveal propos model effect arrhythmia detect signific better exist hierarch network model | D012815;D058067 | T062;T066 |
36,236,494 | smart alcoholmet sensor base deep learn visual percept process autom general enabl enhanc product product qualiti consist alongsid product metric liquor product industri scale also follow autom trend howev small medium produc lag equip modern due high cost industri equip one import sensor autom equip distilleri alcohol concentr sensor use fraction separ process autom supervis paper propos novel low cost approach alcohol concentr sens employ deep learn visual percept tradit alcoholmet purpos train model dataset acquisit apparatus develop larg dataset label imag alcoholmet read acquir problem read alcohol concentr alcoholomet imag treat regress classif problem perform regress classif model investig resnet architectur choic model achiev satisfi perform metric demonstr feasibl propos approach propos system implement raspberri pi camera integr new distil equip addit use retrofit exist equip due non invas natur read scope use expand read type analog instrument simpli retrain model | D000077321;D016571 | T170;T066 |
36,236,492 | improv kinect recognit method identifi unsaf behavior metro passeng order solv problem low action recognit accuraci passeng unsaf behavior caus redund joint studi propos effici recognit method base kinect sensor method use pelvi start point vector high frequenc bone joint end point construct recognit featur vector joint angl differ action obtain use cosin law initi test result convert action similar combin dtw similar algorithm consid combin angl featur joint featur select method combin recognit model form compar experi carri identifi five type unsaf behavior metro passeng punch one arm call help arm help trip forward trip backward result show overal select joint poor similar effect achiev purpos recognit overal recognit model effect local pelvi diverg method higher local adjac joint method local pelvi diverg method highest recognit result maximum angl differ model recognit result five unsaf behavior respect recognit accuraci method indic feasibl model recognit result concentr stabl signific improv recognit rate metro passeng unsaf behavior | D000465;D012919 | T054;T170 |
36,236,491 | accuraci commerci avail fit tracker patient stroke background fit tracker could repres easi use cheap tool continu track physic activ stroke survivor period recoveri home aim studi examin accuraci fitbit activ tracker locomotor activ monitor stroke survivor respect gait disord walk speed walk aid placement tracker bodi method twenti four ambulatori stroke survivor men women locomotiongait disord involv studi patient underw two walk test fitbit alta hr tracker attach differ place bodi accuraci tracker analyz group patient walk without walk aid use singl point stick use roll walker result aid patient accur place waist patient singl point stick reveal smallest deviat tracker attach healthi lower limb patient roll walker reveal smallest deviat tracker attach paret lower limb conclus accuraci compar healthi popul reach three group patient fulfil condit minimum speed kmh optim placement tracker respect walk aid aspect impair | D000072936;D020521 | T047;T073 |
36,236,490 | digit twin hydropon grow bed intellig aquapon system use autom internet thing iot smart technolog rapid introduc develop agricultur technolog sens remot monitor predict tool use purpos enhanc agricultur process aquapon among improv qualiti product digit twin enabl test implement improv physic compon implement comput tool twin virtual environ paper present framework develop digit twin aquapon system framework valid develop digit twin grow bed aquapon system real time monitor paramet name ph electroconduct water temperatur relat humid air temperatur light intens support use artifici intellig techniqu exampl predict growth rate fresh weight grow crop digit twin present base iot technolog databas central control system virtual interfac allow user feedback control system visual state aquapon system real time | D000383;D018556 | T090;T002 |
36,236,489 | flexibl effici secur framework mani mani communic publishsubscrib architectur messag queu telemetri transport mqtt lightweight publishsubscrib protocol current one popular applic protocol internet thing iot thank simplic use scalabl secur version mqtts combin mqtt transport layer secur tls protocol sever shortcom offer one one secur support limit number secur featur high comput communic cost paper propos flexibl lightweight secur solut integr mqtt address mani mani communic reduc communic overhead comput overhead setup secur connect client side | D003142;D013686 | T060;T054;T170;T097;T080 |
36,236,488 | local method ant coloni optim nonuniform space purpos geograph locat select make best use space geograph locat select contain larg amount spatiotempor data constraint result mani solut therefor paper adopt ant coloni algorithm meta heurist search method combin incomplet quadtre improv searchabl space paper propos improv ant coloni algorithm nonuniform space solv p center facil locat problem geograph space divid incomplet quadtre ant coloni path construct level quadtre divis ant coloni leav pheromon multipl search path optim quadtre encod nonuniform space store pheromon matric distanc matric algorithm propos paper improv pheromon diffus algorithm optim object time updat pheromon nonuniform space obtain ideal solut result show algorithm excel perform solv locat problem good converg accuraci calcul time | D000465;D010675 | T170;T123 |
36,236,486 | high sensit flexibl capacit pressur sensor base align airgap dielectr flexibl capacit pressur sensor simpl structur low power consumpt attract attent owe wide rang applic wearabl electron devic howev difficult manufactur pressur sensor high sensit wide detect rang low detect limit develop high sensit flexibl capacit pressur sensor base porous ecoflex align airgap structur manufactur simpli use mold micro needl exist precis align airgap structur signific improv sensor sensit compar dielectr structur without airgap propos capacit pressur sensor align airgap structur support wide rang work pressur kpa quick respons time ms high oper stabil low pressur detect limit pa moreov also studi applic puls wave monitor wearabl sensor exhibit excel perform wearabl devic detect puls wave exercis propos pressur sensor applic electron skin wearabl medic assist devic owe excel function featur | D000076251 | T073 |
36,236,485 | lightweight depth complet network local similar preserv knowledg distil depth percept capabl one essenti requir various autonom drive platform howev accur depth estim real world set still challeng problem due high comput cost paper propos lightweight depth complet network depth percept real world environ effect transfer teacher knowledg use depth complet introduc local similar preserv knowledg distil lspkd allow similar local neighbor transfer distil lspkd lightweight student network precis guid heavi teacher network regardless densiti ground truth data experiment result demonstr method effect reduc comput cost train infer stage achiev superior perform lightweight network | D000465 | T170 |
36,236,484 | concentratenet multi scale object detect model advanc drive assist system use real time distant region locat techniqu paper propos deep learn base object detect method locat distant region imag real time concentr distant object vehicular front camcord perspect tri solv one common problem advanc driver assist system ada applic detect smaller faraway object confid bigger closer object paper present effici multi scale object detect network term concentratenet detect vanish point concentr near distant region initi object detect model inferenc produc larger scale recept field detect result predict potenti vanish point locat farthest locat frame imag crop near vanish point locat process object detect model second inferenc obtain distant object detect result final two inferenc result merg specif non maximum suppress nms method propos network architectur employ object detect model propos model implement state art object detect model check feasibl compar origin model use higher resolut input size concentratenet architectur model use lower resolut input size less model complex achiev signific precis recal improv moreov propos concentratenet architectur model success port onto low power embed system nvidia jetson agx xavier suit real time autonom machin | D001334;D016571 | T056;T170 |
36,236,483 | review multiscal deep learn applic general exist convolut neural network cnn base deep learn model suffer spatial inform loss inadequ featur represent issu due inabl captur multiscal context inform exclus semant inform throughout pool oper earli layer cnn network encod simpl semant represent edg corner latter part cnn network encod complex semant featur complex geometr shape theoret better cnn extract featur differ level semant represent task classif segment work better simpl complex featur map util henc also crucial emb multiscal capabl throughout network various scale featur optim captur repres intend task multiscal represent enabl network fuse low level high level featur restrict recept field enhanc deep model perform main novelti review comprehens novel taxonomi multiscal deep learn method includ detail sever architectur strength implement exist work predomin multiscal approach deep learn network class two categori multiscal featur learn multiscal featur fusion multiscal featur learn refer method deriv featur map examin kernel sever size collect larger rang relev featur predict input imag spatial map multiscal featur fusion use featur differ resolut find pattern short long distanc without deep network addit sever exampl techniqu also discuss accord applic satellit imageri medic imag agricultur industri manufactur system | D000077321 | T066 |
36,236,479 | improv monitor wildlif invas data augment extract append segment entiti owe continu increas damag farm due wild anim destruct crop south korea various method propos resolv issu instal electr fenc use warn lamp ultrason wave recent new method attempt appli deep learn base object detect techniqu robot howev effect train deep learn base object detect model overfit bias train avoid furthermor huge number dataset requir particular establish train dataset specif wild anim requir consider time labor therefor studi propos extract append data augment method specif object extract limit number imag via semant segment correspond object append numer arbitrari background imag thus studi aim improv model detect perform generat rich dataset wild anim various background imag particular imag water deer wild boar current caus problemat social issu comparison object detector train use propos extract append techniqu train use exist data augment techniqu show mean averag precis map improv moreov improv detect perform deep learn base object detect model expect propos techniqu solv issu lack specif data difficult obtain | D000077321;D003670 | T015;T066 |
36,236,478 | locat imag fiber break cfrp use guid wave tomographi eddi current test paper guid lamb wave tomographi eddi current test ect techniqu combin locat evalu fiber break carbon fiber reinforc plastic cfrp structur guid wave test gwt comput tomographi ct imag employ quick locat fiber break cfrp plate b scan perform along two differ fiber orient degre parallel beam project differ featur extract guid wave signal use signal process techniqu wavelet hilbert transform statist function skew kurtosi parallel beam project individu featur use input comput tomographi imag reconstruct approxim estim locat fiber break obtain reconstruct imag imag fusion techniqu appli get complementari inform multipl sourc imag one singl imag locat fiber break c scan perform vicin damag use ect probe doubl excit configur evalu condit fiber break | D002244;D010969 | T121;T167;T196 |
36,236,477 | comput optim imag base reinforc learn robot robot field deepli influenc advent deep learn recent year trend character adopt larg pretrain model robot use case compat comput hardwar avail robot system moreov larg comput intens model imped low latenc execut requir mani close loop control system work propos differ strategi improv comput effici deep learn model adopt reinforc learn rl scenario use case project consid imag base rl method synergi push grasp action first optim step reduc model architectur complex decreas number layer alter architectur structur second consid downscal input resolut reduc comput load final perform weight quantize compar post train quantize quantiz awar train benchmark improv introduc optim run standard test routin show optim strategi introduc improv comput effici around time also slight improv function perform system addit demonstr close loop control behaviour real world robot process everyth jetson xavier nx edg devic | D012371 | T090 |
36,236,476 | deep learn model classif dental diseas use orthopantomographi x ray opg imag teeth challeng materi work human bodi exist method detect teeth problem characteris low effici complex experienti oper higher level user intervent older oral diseas detect approach manual time consum requir dentist examin evalu diseas address concern propos novel approach detect classifi four common teeth problem caviti root canal dental crown broken root canal base deep learn model studi appli yolov deep learn model develop autom tool capabl diagnos classifi dental abnorm dental panoram x ray imag opg due lack dental diseas dataset creat dental x ray dataset detect classifi diseas size dataset use augment imag dataset compris dental panoram imag dental disord caviti root canal bdr dental crown dataset divid train test imag train model yolov evalu test imag train experi demonstr propos model achiev accuraci perform better exist state art model term accuraci univers use dataset model | D000077321;D009057;D014070 | T047;T066;T023 |
36,236,475 | parcemon iot platform real time parcel level last mile deliveri greenhous gas emiss report manag transport australia third largest sourc greenhous gase account around emiss recent time particular result global pandem rapid growth within e commerc sector contribut last mile deliveri becom one main emiss sourc deliveri vehicl oper last mile travel long rout deliv custom array consign parcel vari number weight therefor vehicl play major role increas emiss air pollut work report paper aim address challeng develop iot platform measur report real world last mile deliveri emiss evalu help understand factor contribut freight emiss appropri mitig measur implement unlik previous research complet control laboratori set data collect research deliveri vehicl real world traffic drive condit iot platform test provid contextualis report take account three main context includ vehicl environ drive behaviour approach data collect enabl analysi parcel level emiss correl vehicl characterist road condit ambient temperatur environment factor drive behaviour impact emiss raw data collect sensor analys real time iot platform result show trade parcel weight total distanc travel must consid select best deliveri order reduc emiss overal studi demonstr feasibl iot platform collect desir level data provid detail analysi emiss parcel level type micro level understand provid import knowledg base enhanc deliveri process reduct last mile deliveri emiss | D000393;D000074382 | T131;T197 |
36,236,474 | develop applic compton camera review histori compton camera began detect radiat sourc origin applic astronomi compton camera promis ray detector oper wide energi rang ten kev mev ray detect method compton camera base compton scatter kinemat use determin direct energi ray without use mechan collim although compton camera origin design astrophys applic later appli medic imag well moreov applic environment radiat measur also studi although review paper regard compton camera publish either focus specif detector use camera particular applic compton camera thus aim paper review featur type compton camera introduc applic associ imag algorithm improv scope futur aspect | D000465;D003952 | T170;T060 |
36,236,472 | experiment assess depth estim transpar transluc scene intel realsens sr l rgb camera becom common mani research field sinc inexpens devic provid dens inform observ scene past year realsens rang intel introduc new cost effect rgb sensor differ technolog sophist hardwar softwar model sr l exampl success camera launch intel realsens three camera differ sinc distinct oper principl behavior concern depth estim presenc mani error sourc also specif instanc semi transpar scatter media expect error sourc rgb sensor main new contribut paper full evalu comparison three intel realsens camera scenario transpar transluc propos experiment setup involv aquarium liquid evalu base repeatabilityprecis statist distribut acquir depth allow us compar three camera conclud intel realsens overal best behavior name concern statist variabl also known precis repeat also concern valid measur | D012984 | T090;T170;T073 |
36,236,470 | advanc indoor posit indoor navig locat devic indoor environ becom key issu mani emerg locat base applic intellig space differ field | D000465 | T170 |
36,236,469 | navig path base univers mobil manipul integr control nummic demand servic robot increas mobil manipul robot perform various task dynam environ attract great attent control control mobil platform manipul arm simultan effici perform difficult appli univers sinc base one mobil manipul model lack versatil big problem mobil manipul robot made connect mobil platform manipul differ compani overcom problem paper propos simultan control appli one model also various type mobil manipul robot propos control three main characterist follow establish pose motion plan carri posit avoid obstacl stop stabl manner target coordin prevent robot collis surround obstacl drive defin safeti area manipul hit obstacl drive execut manipul accord control fulli compat robot oper system ros use success three differ type mobil manipul robot addit conduct motion plan experi five target two simul world two motion plan scenario use real robot real world environ result show signific improv time compar exist control method various type mobil manipul demonstr control work success real environ propos control avail github | D012371 | T090 |
36,236,468 | novel reconstruct algorithm high perform compress ultrafast imag compress ultrafast photographi cup type two dimension imag techniqu observ ultrafast process intellig reconstruct method influenc imag qualiti essenti part cup system howev exist reconstruct algorithm most reli imag prior complex paramet space therefor usual take lot time obtain accept reconstruct result limit practic applic cup paper propos novel reconstruct algorithm name pnp ffdnet provid high qualiti high effici compar previous method first built forward model cup three sub optim problem obtain use altern direct multipli method admm close form solut first sub optim problem deriv second inspir pnp admm framework use advanc denois algorithm base neural network name ffdnet solv second sub optim problem real cup data psnr ssim improv averag db respect compar tradit algorithm benchmark dataset real cup data propos method reduc run time averag state art algorithm show compar visual result much shorter run time | D000465;D016571 | T170 |
36,236,467 | activ free user identif use wearabl base vision techniqu order achiev promis smart space environ act fulfil need user unobtrus person manner necessari provid mean seamless continu identif user know inde interact system smart servic provid paper propos new approach capabl perform activ free identif user base hand arm motion pattern obtain wrist worn inerti measur unit imu approach constrain particular type movement gestur activ thus allow user perform freeli unconstrain daili routin user identif take place evalu approach base imu data collect peopl perform daili routin unconstrain result indic approach abl perform activ free user identif accuraci user without requir direct input specif action user furthermor evalu provid evid regard robust approach various differ configur | D000076251;D014953 | T073;T029 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.