mirrors
/
PaddleOCR
mirror of https://github.com/PaddlePaddle/PaddleOCR.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

修改LaTeXOCR的数据处理部分，将生成的数据集中的绝对路径改为相对路径 (#13702) 

* test

* dataprocess_abspath2relpath
pull/13716/head
liuhongen1234567 2024-08-20 15:45:57 +08:00 committed by GitHub
parent 01e60ff9e1
commit 1752c56cb7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
5 changed files with 8 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
configs/rec/rec_latex_ocr.yml
View File

@ -69,6 +69,7 @@ Metric:
Train:
dataset:
name: LaTeXOCRDataSet
data_dir: ./train_data/LaTeXOCR/train
data: ./train_data/LaTeXOCR/latexocr_train.pkl
min_dimensions: [32, 32]
max_dimensions: [672, 192]
@ -99,6 +100,7 @@ Train:
Eval:
dataset:
name: LaTeXOCRDataSet
data_dir: ./train_data/LaTeXOCR/val
data: ./train_data/LaTeXOCR/latexocr_val.pkl
min_dimensions: [32, 32]
max_dimensions: [672, 192]

2
docs/algorithm/formula_recognition/algorithm_rec_latex_ocr.en.md
View File

@ -63,7 +63,7 @@ Evaluation:
# Validation set evaluation
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams
# Test set evaluation
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams Eval.dataset.data=./train_data/LaTeXOCR/latexocr_test.pkl
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams Eval.dataset.data_dir=./train_data/LaTeXOCR/test Eval.dataset.data=./train_data/LaTeXOCR/latexocr_test.pkl
```
Prediction:

2
docs/algorithm/formula_recognition/algorithm_rec_latex_ocr.md
View File

@ -71,7 +71,7 @@ python3 tools/train.py -c configs/rec/rec_latex_ocr.yml -o Global.eval_batch_ste
# 验证集评估
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams
# 测试集评估
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams Eval.dataset.data=./train_data/LaTeXOCR/latexocr_test.pkl
python3 tools/eval.py -c configs/rec/rec_latex_ocr.yml -o Global.pretrained_model=./rec_latex_ocr_train/best_accuracy.pdparams Eval.dataset.data_dir=./train_data/LaTeXOCR/test Eval.dataset.data=./train_data/LaTeXOCR/latexocr_test.pkl
```
### 3.4 预测

4
ppocr/data/latexocr_dataset.py
View File

@ -42,6 +42,7 @@ class LaTeXOCRDataSet(Dataset):
loader_config = config[mode]["loader"]
pkl_path = dataset_config.pop("data")
self.data_dir = dataset_config["data_dir"]
self.min_dimensions = dataset_config.pop("min_dimensions")
self.max_dimensions = dataset_config.pop("max_dimensions")
self.batchsize = dataset_config.pop("batch_size_per_pair")
@ -128,7 +129,8 @@ class LaTeXOCRDataSet(Dataset):
images_transform = []
for img_path in ims:
for file_name in ims:
img_path = os.path.join(self.data_dir, file_name)
data = {
"img_path": img_path,
}

1
ppocr/utils/formula_utils/math_txt2pkl.py
View File

@ -44,6 +44,7 @@ def txt2pickle(images, equations, save_dir):
):
divide_h = math.ceil(height / 16) * 16
divide_w = math.ceil(width / 16) * 16
im = os.path.basename(im)
data[(divide_w, divide_h)].append((eqs[indices[i]], im))
data = dict(data)
with open(save_p, "wb") as file: