Unlimited-OCR 3B
Free.ai (self-hosted)
·
ocr
·
~300 Whakahauhau page
Ko te Unlimited-OCR 3B he a Tauira OCR i hangaia e te Baidu. He kaha rawa i te Image / scan to structured Markdown, complex layouts, tables, multi-page documents.. I whakanohoia ki ngā GPUs Free.ai - e haere ana i te wātea i runga i tō tātau pūpū tohu i ia rā (300 ngā tohu i ia pātaka). I tukua i raro i te MIT - ka whakaaetia te whakamahinga hokohoko i runga i te Free.ai.
Ka whakamahia mā te API
API REST OpenAI-hōia. Whakana tētahi kī, ā, ka karangatia tēnei tauira i roto i ngā takirua.
curl -X POST https://api.free.ai/v1/ocr/ \
-H "Authorization: Bearer sk-free-..." \
-H "Content-Type: application/json" \
-d '{"model":"unlimited-ocr","prompt":"your prompt here"}'He pātai pānui
Ka tangohia e te Unlimited-OCR 3B te kupu mai i ngā whakaahua me ngā PDFs. Whakarewa i tētahi pātaka kātaki, tētahi whakaahua, tētahi whakaahua rānei o tētahi tuhinga, ā, ka hoki mai te Unlimited-OCR 3B i te kupu - e tiaki ana i ngā whakawhitinga raina me (i runga i ngā mīhini kaha) ngā ripanga, ngā tātai, me te hanganga.
E whakahaere ana a Unlimited-OCR 3B i ngā taumaha o ngā tuhipānui - Latin, Cyrillic, CJK, Arabic, Devanagari, me ētahi atu. He rārangi reo katoa te /ocr/ landing mō ia mīhini.
Heoi mō te tuhituhi ā-ringa mārama; He uaua ake te āhua o te whakahua me te whakahua. TrOCR me ngā mīhini whakarerekē ā-hōu, he tino pai ake i te Tesseract o mua i runga i te tuhituhi.
Ko ngā mīhini mōhio-whakahaere (GOT-OCR, Marker, Florence-2) ka hoki ki ngā ripanga Markdown/HTML me LaTeX mō ngā tātai. Ka hoki ngā mīhini kupu-pūnoa ki te kupu kāore i te hanganga - tirohia te āhuatanga huaputa o tēnei tauira.
Ka haere a Unlimited-OCR 3B ki a tātau GPUs mai i tōtou pū wairangi wātea i ia rā; $5 → 200,000 ngā tohu whakawhiwhita i muri mai. ~300 ngā tohu i ia pātaka.
PNG, JPG, WebP, HEIC, BMP, tae atu ki ngā PDF kotahi me ngā pātaka maha tae atu ki te 50 MB. Ka tika te whakarerekētanga, te whakarerekētanga rānei i mua i te whakawāteatanga.
Ko te tika o te āhua o te āhua o te āhua o te>98% i runga i ngā pātaka tātuhi mārō, e whakarewa ana i ngā whakaahua iti-resolution, ngā karapa tino pōhēhē, ngā tātai ā-kore rānei. Tērā ngā mīhini whakataurite i runga i te /ocr/compare/ ina pā te tika.
He - /batch/ e whakaae ana ki tētahi pūrākau o ngā whakaahua/PDFs. Ka tū ia whakawāteatanga ki /account/?tab=hitori. Mō te tiaki i te rākau pūrākau, ka whakamahia te API.
He - POST tō tō tau te tāuru ki /v1/ocr/ me te tauira "Unlimited-OCR 3B". Ka hoki mai te JSON me te kupu + (i te wā e tautokona ana) te hoahoa + ngā tāruatanga taumata-wānanga. He tohutoro katoa te /api/.
Ko ngā tauira ā-whāinga e pupuri ana i tō tātou pūranga i runga i a tātau GPUs; ka haere tonu te utu mā te DPA. Ka whakakoretia ngā tuhinga i muri i te matapihi tiritiri. Kāore e whakaakona e tātau i o tātau tāurutanga.
He - ka whakaaetia e te Free.ai te whakamahinga hokohoko o te kupu whakawātea. E hiahiatia ana e koe ngā mana ki te tuhinga pūtake.
1-5 ngā takia pātaka. Ka roa ake te wā o ngā mīhini mōhio-whakapae (taupānga, pāngarau). He āhua paerangi te tauine o ngā PDFs maha-pātaka.