PaddlePaddle / PaddlePaddle/FastDeploy
db++通过fastdeploy 无法正常部署
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 3.7k
- Forks
- 756
- Avg merge
- 19h 28m
- Merged PRs (30d)
- 4
Description
环境
-
【FastDeploy版本】: 说明具体的版本,如fastdeploy-'1.0.5'
-
【编译命令】如果您是自行编译的FastDeploy,请说明您的编译方式(参数命令)
-
【系统平台】: / Windows x64(Windows10)
-
【编译语言】: Python(3.10)
问题日志及出现问题的操作流程
在部署pp_ocr_v3的过程中,我训练了一个基于det_mv3_db的检测模型,但由于精度有要求,因此我基于r50_db++重新训练了检测模型,但在使用fd部署的时候,基于det_mv3_db的检测模型可以正确的检测出我需要的内容,而db++模型无法正常检测。
`
import fastdeploy.vision as vision
mv3_db
det_model = vision.ocr.DBDetector(r"inference_model/db/inference.pdmodel",r"inference_model/db/inference.pdiparams")
db++
det_model = vision.ocr.DBDetector(r"inference_model/det_db/inference_.pdmodel",r"inference_model/det_db/inference_.pdiparams")
rec_model = vision.ocr.Recognizer("inference_model/rec_ocrv3/inference.pdmodel",
"inference_model/rec_ocrv3/inference.pdiparams",
"inference_model/rec_ocrv3/num.txt",
)
ppocr_v3 = fd.vision.ocr.PPOCRv3(det_model=det_model,cls_model=None,rec_model=rec_model)
`
db++检测结果:
print(result)
rec text: 4 rec score:0.999919
mv3_db检测结果
print(result)
det boxes: [[99,32],[309,30],[310,118],[100,119]]rec text: 4 rec score:0.999997
但是我通过命令行直接推理的能够得出正常结果
python3 tools/infer/predict_det.py --image_dir="/home/aistudio/data/images_split_paddle/ae90bb62-1185-470d-87d8-341c7c5c8ee8__33.jpg" --det_model_dir="./inference/det_db++/" --det_algorithm="DB++"
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by comparing the FastDeploy DBDetector setup with the command-line path in tools/infer/predict_det.py, using the provided DB and DB++ model directories and the DB++ algorithm setting. Reproduce the differing detection results on the referenced image, then identify whether FastDeploy supports the DB++ model configuration and add a regression test or documentation once the expected output is defined.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100