PaddleOCR
Free.ai (self-hosted)
·
ocr
·
~300 Ama-token ngalinye page
PaddleOCR yi an Imodeli ye-OCR eyenziwe ngu PaddlePaddle. Onamandla kakhulu ku Multi-language OCR, document layout extraction, receipts.. I-self-hosted ku-Free.ai GPUs - isebenza ngokukhululekile ngaphezu kwe-token pool yakho yansuku zonke (300 tokens Ikhasi ngalinye). Kukhishwa ngaphansi kwe-Apache 2.0 — ukusetshenziswa kokuthengiswa kwempahla kuvunyelwe ku-Free.ai.
Sebenzisa nge-API
I-REST API ehambisana ne-OpenAI. Dala isithonjana bese ubiza le modeli emaminithini.
curl -X POST https://api.free.ai/v1/ocr/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"model":"paddleocr","prompt":"your prompt here"}'Imibuzo ebuzwa kaningi
PaddleOCR isusa umbhalo kusuka kuzithombe ne-PDFs. Layisha phezulu ikhasi eliscanwe, isithombe-skrini, noma isithombe sedokhumende futhi PaddleOCR ibuyisela umbhalo — igcina amaphutha omgwaqo kanye (ezindaweni ezikwaziyo) nethebula, ifomu, nesakhiwo.
PaddleOCR iphatha ama-scripts angaphezu kwekhulu — isiLatini, isiCyrillic, iCJK, isiArabic, iDevanagari, nezinye eziningi. I /ocr/ landing ine-list ephelele ye-language ngenjini ngayinye.
Yebo, ukubhala ngesandla okuphrintiwe okuhlanzekile; i-cursive ne-shorthand kunzima. TrOCR kanye ne-transformer-based engines yamanje ingcono kakhulu kune-Tesseract ekhona yokubhala ngesandla.
Izinjini ezizizwayo zesitayela (GOT-OCR, Umqondisi, Florence-2) zibuyisela amathebula weMarkdown/HTML neLaTeX ngefomula. Izinjini zombhalo ocacile zibuyisela umbhalo ongenasakhiwo — hlola indlela yokuphuma yale modeli.
PaddleOCR isebenza kuma-GPU ethu kusuka epulazini lakho elimahhala ligcwele; $5 → 200,000 ama-token akhokhelwe ngemuva. ~300 ama-token ngekhasi.
[Translation temporarily unavailable. Please try again.]
Uhlaka lokucacisa uphawu luvame ukuba>98% kumakhasi aphrintiwe ahlanzekile, ukubopha kuma-photos ane-resolute ephansi, ukuskena okubi kakhulu, noma amafonti angajwayelekile. Qaphela ama-engine ku /ocr/compare/ uma uhlaka lubalulekile.
Yebo — /batch/ i accetta ifolda yezithombe/PDFs. Ukukhipha ngakunye ku /account/?tab=history. Ukugcinwa kwefolda-yesihlahla sebenzisa i-API.
Yebo — IKHAYA lakho lithunyelwe ku /v1/ocr/ ngemodeli "PaddleOCR". Libuyisela i-JSON ngetekisi + (uma kuxhaswe) isitayela + amabhokisi okungenagama. /api/ anesiqalo esigcwele.
Amamodeli ahlala ahlala agcina ifayela lakho ku-GPUs ethu; ama-premium adlula nge-DPA. Amadokhumende asuswa ngemuva kokuhlukaniswa kwefasitela. Asiqeqeshi ngemingeniso yakho.
Yebo — Free.ai inikeza ukusetshenziswa komnotho kwetekisi elikhishwayo. Kudingeka uvumeleke kudokhumende lomsuka.
1–5 imizuzwana ngayinye ikhasi ngalinye. Izinjini ezizizwayo zesitayela (ithebula, i-mathematics) zithatha isikhathi eside. I-PDF enekhasi elilodwa noma eliningi lilinganiselwa ngokuqondile.