isiZulu
Izitimela Ezizihambelayo Nezibonwa Nge-Camera: Izinhlelo Zokubona Ziguqula Kanjani Ukulima Okunembayo
Kwadalwa ngo 01.08
Imboni yezolimo emhlabeni wonke isemgwaqweni. Ngokukhula kwabantu okubikezelwa ukuthi kuzofinyelela izigidi ezingu-9.7 ngo-2050, abalimi banomfutho omkhulu wokukhulisa ukukhiqiza ngenkathi behlisa ukuchitha izinsiza, izindleko zabasebenzi, nomthelela emvelweni. Kungenwa izitimela ezizihambelayo—imishini ezihambela yona engasaseyona i-science fiction kodwa isixazululo esingokoqobo izinseleleni zezolimo zesimanje. Ezingqimeni zalokhu kucwenga kobuchwepheshe kukhona ingxenye ebalulekile: izinhlelo zokubona ngekhamera. Ngokungafani nezinzwa zakudala ezisebenzisa i-radar noma i-LiDAR (ezingabiza kakhulu emapulazini amaningi), ukubona ngekhamera kunikeza indlela esebenza kahle, enokulungiswa okuphezulu evumela izitimela ezizihambelayo ukuthi "zibone" futhi zisebenzisane nemvelo yazo ngokunemba okumangalisayo. Kulesi sihloko, sizohlola ukuthi kanjani izinhlelo zokubona ngekhamera zibuyekeza amandla amata'kthila azihambelayo, zihlukanisa ukuthuthukiswa kwawo, izinhlelo zokusebenza emhlabeni wangempela, izinto ezintsha zobuchwepheshe, kanye nekusasa lokulima okusekelwa ukubona.
Kungani Ukubona Nge-khamera KunguQhawe Ongaziswa Lamata'kthila Azihambelayo
Lapho kuxoxwa ngezimoto ezizihambelayo, i-LiDAR ne-radar kuvame ukuthatha isiteji. Lezi zinzwa zikwazi kakhulu ukuthola izinto nokukala ibanga ezimweni zezulu ezinzima, okwenza zibe yisidingo ezimotweni ezizihambelayo emigwaqweni ematasa. Kodwa-ke, ezolimo zisebenza endaweni ehlukile kakhulu—lapho ukunemba, ukonga izindleko, kanye nekhono lokuhlukanisa phakathi kwezinguquko ezicashile ezitshalweni, enhlabathini, nasezindaweni ezihlukahlukene kubaluleke kakhulu. Kulapho izinhlelo zokubona ngekhamera zikhanya khona.
Izinhlelo zokubona ngekhamera—ezihlanganisa amakhamera aphakeme (HD), izihleli zezithombe, kanye ne-machine learning algorithms—zithwebula idatha ebonakalayo ye-2D neye-3D engahlaziywa ngesikhathi sangempela. Uma kuqhathaniswa ne-LiDAR, engabiza izinkulungwane zamadola, amamojuli ekhamera ayabiza kakhulu, okwenza ubuchwepheshe obuzenzakalelayo bufikeleleke ezindaweni ezincane nezimaphakathi. Ngaphezu kwalokho, amakhamera ahlinzeka ngemininingwane ebonakalayo eyengeziwe: angakwazi ukukhomba umbala weqabunga lesitshalo (okubonisa impilo noma ukushoda kwezakhamzimba), athole ukwakheka komhlabathi (okuqondisa ukujula kokuhlwanyela), futhi akwazi ngisho nokuhlukanisa phakathi kwezitshalo nezilwelwe—imisebenzi i-LiDAR enobunzima ngayo ngenxa yokulinganiselwa kwayo ekuthwebuleni imininingwane emihle.
Elinye inzuzo enkulu yekhamera yokubona ukuthi iyazivumelanisa nezimo. Izinhlelo zanamuhla zisebenzisa amamodeli okufunda komshini aqeqeshwe ezilimo ezithile (isb., ummbila, ukolweni, ubhontshisi wesoya) noma izindawo zokulima (isb., izindawo ezomile, amapulazi amanzi), okwenza izitshalo zikwazi ukusebenza kahle ezimeni ezahlukene. Lokhu kuvumelana nezimo kubalulekile kwezolimo, lapho kungabikho amapulazi amabili afanayo. Ngenxa yalokho, ikhamera yokubona isibe umgogodla wezinhlelo eziningi zezitshala ezizimele, ezizivumela ukuthi zenze imisebenzi eyinkimbinkimbi ngokuphazamiseka okuncane komuntu.
Ukuvela Kokubona Nge-Khamera Kumaloli Azihambelayo: Kusukela Ekutholeni Okuyisisekelo Kuye Ekwenzeni Izinqumo Ezihlakaniphile
Ukubona nge-khamera akulutsha ezolimo—iminyaka eminingi, abalimi bebelokhu besebenzisa amakhamera ayisisekelo ukuze babheke noma baqaphe izitshalo kalula. Kodwa-ke, ukuhlanganisa ukubona nge-khamera kumaloli azihambelayo kubonisa ukwanda okukhulu emakhonweni, okushukunyiswa yintuthuko ku-AI nasekuqhubeni okusezingeni eliphezulu. Ake silandele le nqubekela phambili:
1. Izinyathelo Zokuqala: Ukuthola Izinto Okuyisisekelo
Isizukulwane sokuqala samaloli anamakhamera agxile ekutholeni izinto eziyisisekelo. Lezi zinhlelo zazisebenzisa izindlela ezisekelwe emithethweni ukukhomba izithiyo ezinkulu, njengemithi, amatshe, noma olunye uhlelo lokulima. Nakuba zazisebenza, zazinezinkinga: zazingenakukwazi ukuhlukanisa phakathi kwezinto ezingalimazi (isibonelo, igatsha eliwile) nezibalulekile (isibonelo, izilwane), futhi zazisebenza kabi ekukhanyeni okuncane noma ezimeni zezulu ezimbi (isibonelo, imvula, inkungu).
2. Izinyathelo Eziphakathi: Ukuqashelwa Okunikezwe Amandla Emishini Yokufunda
Ukwethulwa komshini wokufunda (ML) kushintshe izinhlelo zokubona ngekhamera. Ngokuqeqesha amamodeli e-ML ezinkulungwaneni zezithombe zezitshalo, ukhula, inhlabathi, nezithiyo, abathuthukisi banikeze amatafulela amandla okubona nokuhlukanisa izinto ngokunemba okuphezulu. Ngokwesibonelo, itafulela elihlome ngekhamera enikwe amandla yi-ML lingahlukanisa phakathi kwesithombo sikakolweni nokhula, okuvumela ukufafaza izibulala-zinambuzane okuhlosiwe—kunciphisa ukusetshenziswa kwamakhemikhali kuze kufike ku-90% uma kuqhathaniswa nokufafaza okubanzi. Lesi sigaba sabona futhi ukwamukelwa kwamakhamera e-stereo, asebenzisa amalensi amabili ukuthwebula ulwazi lokuqhakama kwe-3D, okuvumela amatafulela ukuthi azulazule ezindaweni ezingalingani futhi alungise ukuphakama kwawo noma isivinini ngokufanele.
3. Isimo Samanje: Ukwenza Izinqumo Ezihlakaniphile Ngesikhathi Sangempela
Amathangi azihambelayo anamuhla athuthuke kakhulu anezinhlelo zokubona ngekhamera ezihlanganiswe ne-edge computing kanye ne-deep learning. I-Edge computing ivumela ukucubungula kwedatha kusithangi, iqeda ukubambezeleka okuhambisana nokucubungula okusekelwe emafini—ikhono elibalulekile emisebenzini edinga izinqumo ezenziwa ngokushesha, njengokugwema izithiyo ezingalindelekile noma ukulungisa ubuningi bokuhlwanyela. Amamodeli e-deep learning, njenge-convolutional neural networks (CNNs), avumela amathangi ukuthi ahlaziye idatha eyinkimbinkimbi ebonakalayo ngesikhathi sangempela: angakwazi ukuthola izimpawu zokuqala zezifo zesivuno (isibonelo, amaqabunga kakolweni aphuzi), aqaphe amazinga omswakama wobumba ngokuhlaziya imibala, futhi angakwazi ngisho nokubikezela izivuno ngokusekelwe empilweni yezitshalo.
Esinye isibonelo esivelele yi-John Deere’s 8R Autonomous Tractor, ehlanganisa amakhamera nezinye izinzwa ukwenza imisebenzi yokulima, ukuhlwanyela nokuvuna. Uhlelo lwayo lokubona ngekhamera lungakwazi ukubona imingcele yenkambu ngokunemba okungaphansi kwesentimitha, luqinisekise ukuthi itrektara ihlala endaweni emisiwe futhi igwema ukudlula izindawo ezifanayo—kunciphisa ukuchitha uphethiloli nokuthuthukisa ukusebenza kahle.
Izicelo Ezingokoqobo: Indlela Ukubona Ngekhamera Okuguqula Ngayo Imisebenzi Yelima
Izinhlelo zokubona ngekhamera azilona nje ubuchwepheshe obuhle—ziletha imiphumela ebonakalayo kubalimi emhlabeni wonke. Ngezansi kunezinhlelo ezithinta kakhulu ezisebenzisa amata'kthila azihambelayo anamakhamera:
1. Ukuhlwanyela nokutshala okunembayo
Ukuhlanywa yisikhathi esibalulekile esithinta ngqo imiphumela yezitshalo. Izindlela zokuhlanywa zakudala zivame ukubangela ukusatshalaliswa okungalingani kwembewu noma ukungafani kokujula kokutshala, okuholela ekuhlanyweni okungaphumeleli. Amaloli azisebenzelayo ahlome ngezinhlelo zokubona ngekhamera axazulula le nkinga ngokuhlaziya izimo zenhlabathi ngesikhathi sangempela. Amakhamera athatha izithombe zokwakheka kwenhlabathi nokuqukethwe umswakama, futhi izinhlelo ze-AI zinquma ukujula okufanele kokuhlanywa nokuhlukaniswa kwembewu ngayinye. Ngokwesibonelo, enhlabathini eyomile, iloli litshala imbewu ujule ukuze ifinyelele umswakama, kuyilapho enhlabathini emanzi, litshala kancane ukuze ligweme ukugcwala kwamanzi. Lokhu kunemba kungandisa amazinga okuhlanywa kufike ku-20%, kukhulise imiphumela yezitshalo iyonke.
2. Ukulawulwa Kwezinambuzane Nezihlungu
Ukhula nezinambuzane zibeka izinsongo ezinkulu empilweni yezitshalo, kodwa izindlela zokulawula zakudala (isibonelo, ukufafaza okubanzi izibulala-zinhlwa, ukufafaza okubanzi izibulala-zinambuzane) azisebenzi kahle futhi ziyingozi emvelweni. Umbono wekhamera uvumela amatafula azihambelayo ukuthi enze "ukwelashwa okuyindawo": amakhamera ayascan insimu, athole ukhula noma izitshalo ezihlaselwe izinambuzane, futhi aqondise isifafazi setafula ukuthi sifafaze amakhemikhali endaweni ezithintekayo kuphela. Lokhu akunciphisi ukusetshenziswa kwamakhemikhali kuphela kodwa futhi kunciphisa umonakalo ezinambuzaneni ezizuzisayo kanye namagciwane omhlabathi. Ucwaningo lweNyuvesi yaseCalifornia lwathola ukuthi ukufafaza okuyindawo okuqondiswa yikhamera kunciphise ukusetshenziswa kwezibulala-zinhlwa ngo-75% ngenkathi kugcinwa izinga elifanayo lokulawula ukhula njengokufafaza okubanzi.
3. Ukuqapha Impilo Yezitshalo kanye Nokubikezela Isivuno
Ukuqashelwa kusenesikhathi kwezifo zezilimo nokushodelwa kwezakhamzimba kubalulekile ekunciphiseni ukulahleka kwesivuno. Izinhlelo zokubona ngekhamera ezihlomele amakhamera e-multispectral—azithwebula ukukhanya ngaphezu kwe-spectrum ebonakalayo—zingaqashela izinguquko ezicashile empilweni yezilimo ezingabonakali ngamehlo omuntu. Ngokwesibonelo, izithombe ze-near-infrared (NIR) zingabonisa ukucindezeleka kwamanzi ezitshalweni, kanti izithombe ze-red-edge band zingabonisa ukushodelwa kwe-nitrogen. Amaloli azihambelayo angahlola amasimu njalo, athwebule futhi ahlaziye lezi zithombe, futhi azise umlimi ngezinkinga ezingenzeka ngaphambi kokuba zisabalale. Ezinye izinhlelo ezithuthukile zisebenzisa ngisho ne-AI ukubikezela izivuno zezilimo ngokusekelwe kudatha ebonakalayo, kusiza abalimi ukuthi benze izinqumo ezinolwazi mayelana namasu okuvuna nokumaketha.
4. Ukuzulazula Okuzimele Nokuthola Imingcele
Ukuqondisa ensimini yepulazi kunzima kuneziqondiso zomgwaqo omkhulu—izinkundla zivame ukuba nesimo esingajwayelekile, nezingqinamba ezifana nezihlahla, izinga, nezinhlelo zokunisela. Uma zihlanganiswa ne-GPS, izinhlelo zokubona ngekhamera zinika amandla ama-tractor azihambelayo ukuba aqondise lezi zinkinga ngokunemba. Amakhamera athwebula izithombe zemingcele yenkundla, futhi ama-algorithm e-AI asebenzisa le datha ukudala imephu yesikhathi sangempela yendawo. I-tractor ingabe isilungisa indlela yayo ukugwema izithiyo nokuhlala ngaphakathi kwemingcele yenkundla, iqinisekisa ukuthi zonke izintshi zenkundla ziyamboziwe ngaphandle kokudlula izikhathi eziningi endaweni efanayo. Lokhu akugcini ngokonga isikhathi negesi kodwa futhi kunciphisa ukuminywa komhlabathi—inqaba enkulu kwezolimo zesimanje—ngokunciphisa inani lezikhathi i-tractor ehamba phezu kwendawo efanayo.
Ukuxazulula Izinselele: Ikusasa Lemibono Yamakhamera Kumaloli Azimele
Nakuba umbono wekhamera uthuthuke kakhulu, kusenezinselelo okufanele zinqotshwe ngaphambi kokuba ube yinto ejwayelekile kumaloli azihambelayo. Enye yezinselelo ezinkulu yizimo zezulu ezimbi: imvula, inkungu, uthuli, nokukhanya okuphansi kungalimaza ikhwalithi yesithombe, kube nomthelela ekunembeni kwezinhlelo zokubona ngekhamera. Ukubhekana nalokhu, abathuthukisi basebenza kumalulwazi athuthukisiwe wokuthuthukisa izithombe ahlunga umsindo futhi athuthukise ukubonakala ezimweni ezinzima. Baphinde bahlanganise ukubona ngekhamera nezinye izinzwa (isibonelo, i-LiDAR, i-radar) ngendlela "yokuhlanganisa izinzwa", behlanganisa amandla enye nenye inzwa ukuqinisekisa ukusebenza okuthembekile kuzo zonke izimo.
Enye inselele ukufunwa kwamaqoqo amakhulu, anekhwalithi ephezulu ukuqeqesha amamodeli e-AI. Izitshalo ezahlukene, inhlabathi, nezimo zezulu zidinga idatha yokuqeqesha ehlukile, engathatha isikhathi futhi ibize ukuqoqa. Ukubhekana nalokhu, izinkampani zisebenzisana nabalimi emhlabeni wonke ukwakha amaqoqo ahlukahlukene. Ziphinde zisebenzise idatha eyenziwe (izithombe ezidaliwe ngekhompyutha) ukwengeza idatha yomhlaba wangempela, okuzivumela ukuqeqesha amamodeli ezimo ezingavamile noma eziyisibonelo ngaphandle kokuqoqa izithombe zangempela.
Sibheke phambili, amandla emibono yamakhamera kumaloli azimele awapheli. Ngezansi kunezindlela eziyinhloko okufanele uzibheke:
• Ukwanda kwe-Edge AI: Njengoba ubuchwepheshe be-edge computing buya ngokuya buqina futhi bube nemali ephansi, amaloli azihambelayo azokwazi ukucubungula idatha eyinkimbinkimbi kakhulu ngesikhathi sangempela, okuzovumela imisebenzi eyengeziwe njengokwenza imephu yenzuzo ngesikhathi sangempela nokuphathwa kwezitshalo okuguquguqukayo.
• Ukuhlanganiswa kwe-5G: Ubuchwepheshe be-5G buzovumela ukuxhumana okungenamaphutha phakathi kwamaloli azihambelayo, izinhlelo zokuphatha ipulazi, kanye nefu. Lokhu kuzovumela amaloli ukuthi ahlanganyele idatha ebonakalayo nezinye izinjini nabalimi ngesikhathi sangempela, okuzovumela imisebenzi yokulima ehlangene (isibonelo, iqembu lamaloli elisebenza ndawonye ukutshala insimu).
• I-AI ECHAZISEKAYO: Njengoba izinhlelo ze-AI ziba nzima kakhulu, kunesidingo esikhulayo se-"AI echazekayo"—izinhlelo ezingachaza izinqumo zazo kubalimi. Ngokwesibonelo, uma iloli lithola isifo sezilimo, ngeke nje lixwayise umlimi kodwa futhi lizohlinzeka ngokuhlaziywa okunemininingwane kwezimpawu ezibonakalayo ezazisebenzisa ukwenza leso sinqumo. Lokhu kuzosiza abalimi ukuthi bathembe futhi bamukele ubuchwepheshe.
• Ukubiza ngobunye: Njengoba ubuchwepheshe bekhamera nama-chip e-AI buya ngobiza, izinhlelo zokubona ngekhamera zizotholakala emapulazini amancane kakhulu. Lokhu kuzokwenza ukuthi ukulima okuzenzakalelayo kufinyeleleke kubo bonke, kuvumele abalimi bazo zonke izinhlobo ukuthi bazuze ekukhiqizeni okwandisiwe nasezindlekweni ezincishisiwe.
Isiphetho: Ukubona ngekhamera kuyikusasa lokulima okuzenzakalelayo
Izilimo ezizihambelayo ziguqula imboni yezolimo, futhi izinhlelo zokubona ngekhamera zisemqoka kule ngunguquko. Ngokunikeza izindlela ezingabizi, ezinokulungiswa okuphezulu ukuze izilimo "zibone" futhi zisebenze nemvelo yazo, ukubona ngekhamera kunika amandla izindlela zokulima okunembayo ezazingenakwenzeka ngaphambili. Kusukela ekuhlwanyeleni okunembayo nasekulawuleni ukhula kuya ekugadeni impilo yezitshalo nokuzulazula okuzenzakalelayo, ukubona ngekhamera kusiza abalimi ukuthi bandise umkhiqizo, banciphise ukuchitha izinsiza, futhi bahlangabezane nesidingo esikhulayo somhlaba wonke sokudla.
Nakuba izinselelo zisekhona, ikusasa lokubona ngekhamera kumaloli azihambelayo lithembisa kakhulu. Ngokuqhubeka kobuchwepheshe be-AI, i-edge computing, kanye nokuhlanganiswa kwezinzwa, izinhlelo zokubona ngekhamera zizoba nezinga eliphezulu kakhulu lokwethenjwa kanye namandla, okwenza ukulima okuzihambelayo kufinyeleleke kubalimi emhlabeni wonke. Njengoba sibheka ikusasa elizinzile nelikhiqizayo kwezolimo, into eyodwa icacile: ukubona ngekhamera akuyona nje ingxenye yamaloli azihambelayo—yiso esizohola imboni phambili. Uma ungumlimi ofuna ukwamukela ubuchwepheshe obuzihambelayo noma umthandi wobuchwepheshe onentshisekelo ngekusasa kwezolimo, manje yisikhathi sokuhlola amathuba amaloli azihambelayo anezinhlelo zokubona ngekhamera. Uguquko selufikile, futhi luholwa amandla okubona.
Uxhumane
Sicela uxhumane nathi uhambele
WhatsApp
WeChat