feat: add glm ocr

This commit is contained in:
liuyuanchuang
2026-02-06 15:06:50 +08:00
parent c372a4afbe
commit f0ad0a4c77
5 changed files with 133 additions and 32 deletions

View File

@@ -2,11 +2,11 @@
from fastapi import APIRouter, Depends, HTTPException
from app.core.dependencies import get_image_processor, get_layout_detector, get_ocr_service, get_mineru_ocr_service
from app.core.dependencies import get_image_processor, get_layout_detector, get_ocr_service, get_mineru_ocr_service, get_glmocr_service
from app.schemas.image import ImageOCRRequest, ImageOCRResponse
from app.services.image_processor import ImageProcessor
from app.services.layout_detector import LayoutDetector
from app.services.ocr_service import OCRService, MineruOCRService
from app.services.ocr_service import OCRService, MineruOCRService, GLMOCRService
router = APIRouter()
@@ -18,6 +18,7 @@ async def process_image_ocr(
layout_detector: LayoutDetector = Depends(get_layout_detector),
mineru_service: MineruOCRService = Depends(get_mineru_ocr_service),
paddle_service: OCRService = Depends(get_ocr_service),
glmocr_service: GLMOCRService = Depends(get_glmocr_service),
) -> ImageOCRResponse:
"""Process an image and extract content as LaTeX, Markdown, and MathML.
@@ -33,18 +34,16 @@ async def process_image_ocr(
Use the /convert/latex-to-omml endpoint to convert LaTeX to OMML separately.
"""
image = image_processor.preprocess(
image_url=request.image_url,
image_base64=request.image_base64,
)
try:
if request.model_name == "mineru":
image = image_processor.preprocess(
image_url=request.image_url,
image_base64=request.image_base64,
)
layout_info = layout_detector.detect(image)
if layout_info.MixedRecognition:
ocr_result = mineru_service.recognize(image)
elif request.model_name == "paddle":
ocr_result = paddle_service.recognize(image)
else:
raise HTTPException(status_code=400, detail="Invalid model name")
ocr_result = glmocr_service.recognize(image)
except RuntimeError as e:
raise HTTPException(status_code=503, detail=str(e))