GOT-OCR v2
Free.ai
·
ocr
·
~500 tokens per page
GOT-OCR v2 waa a Midab OCR. Waddooyinka ka hor inta aan qaababka dibadda - ~500 calaamadaha bogga (50% markup ka badan qiimaha upstream).
Ku isticmaal API
OpenAI-ku habboon REST API. abuuro fure iyo wicitaan noocan ah daqiiqado.
curl -X POST https://api.free.ai/v1/ocr/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"model":"premium/got-ocr/v2","prompt":"your prompt here"}'Su'aalaha badanaa la isweydiiyo
GOT-OCR v2 soo saaro qoraalka ka sawirro iyo PDFs. Upload page scanned, screenshot, ama sawirka ah oo ah warqad iyo GOT-OCR v2 soo celin qoraalka - ilaalinta line breaks iyo (oo ku saabsan mashiinada awood) tabobarka, formulas, iyo dhismaha.
GOT-OCR v2 xakamayn karto boqolaal qoraalada - Latin, Cyrillic, CJK, Carabiga, Devanagari, iyo in ka badan. The /ocr/ duulimaadku wuxuu leeyahay liiska luqadda oo dhan ee mashiinka.
Haa, si ay u nadiif ah daabacay-style qoraalka gacanta; cursive iyo shorthand waa ka adag tahay. TrOCR iyo matoorada transformer-ku salaysan casriga ah ayaa si muuqata ka fiican ka hor taariikhda Tesseract on qoraalka gacanta.
Layout-aware mashiinada (GOT-OCR, Marker, Florence-2) soo celin Markdown / HTML table iyo LaTeX for formulas. Plain-text mashiinada soo celin qoraalka aan dhisme - hubi qaabka soo saarka ee noocan ah.
GOT-OCR v2 waa mashiinka OCR premium. About ~ 500-1,500 calaamadaha bogga, muujiyay nool ka hor inta aadan abuurin. Self-hosted OCR ku socdaan bilaash ah ee aad 30,000-token maalin kasta; moodooyinka premium waa lacag-sida-aad-go, la top-ups ka $ 1.
PNG, JPG, WebP, HEIC, BMP, iyo sidoo kale PDF-yada bog kasta iyo bog kasta ilaa 50 MB. Bogga la qalajiyey ama la wareegay ayaa si otomaatig ah loo hagaajin doonaa ka hor soo saarista.
Calaamadaha saxnaanta waa caadi ahaan> 98% on boggaga daabacan nadiif ah, hoos u dhaca on low-res sawirro, scans aad u qalalan, ama fonts aan caadi ahayn. La barbar dhigo mashiinada on /ocr/compare/ marka saxnaanta ay muhiim u tahay.
Haa - / batch / aqbalaa folder of images / PDFs. Dhammaan soosaarka dhulka ku / account /? tab = taariikhda. Si aad u folder-geed ilaalinta isticmaal API.
Haa - POST faylka aad /v1/ocr / la model="GOT-OCR v2". JSON la soo celin doonaa qoraalka + (marka la taageerayo) qaabeynta + bogagga booqashada eray-tayada. /api/ waxaa ka mid ah soo jeedinta buuxda.
Self-hosted qaabab ku hayn faylka ku GPUs our; premium ka gudbaan iyada oo DPA. Documents waa la tirtiri doonaa ka dib share-window. Waxaanan tababarka ku saabsan soo gelitaankaaga.
Haa - Free.ai siin isticmaalka ganacsi ee qoraalka soo bixi. Waxaad u baahan tahay xuquuqda in ay xuquuqda asalka ah.
1-5 ilbiriqsi oo bogga ah caadi ahaan. Layout-aware mashiinada (taallada, xisaabta) qaadataa waqti dheer. Multi-page PDFs scale sicir-dhimista sicir-dhimista.