search

JiaquanYe / TableMASTER-mmocr

2nd solution of ICDAR 2021 Competition on Scientific Literature Parsing, Task B.
Apache License 2.0
442 stars 104 forks source link

questions about pubtabnet #2

Open myh12138 opened 3 years ago

myh12138 commented 3 years ago

after run python ./table_recognition/data_preprocess.py

then, i want to train it sh ./table_recognition/table_text_line_recognition_dist_train.sh but in ./configs/textrecog/master/masterv2_ResnetExtra_tableRec_dataset_dynamic_mmfp16.py alphabet_file = '/data_0/pubtabnet/processed_data/recognitionAlphabet.txt' however the processed data didn't exist recognitionAlphabet.txt' and other files that mentioned in xxxxmmfp16.py. Also, files mentioned in ohter py files didn't satisfied to the data_preprocess.py file

JiaquanYe commented 3 years ago

after run python ./table_recognition/data_preprocess.py

then, i want to train it sh ./table_recognition/table_text_line_recognition_dist_train.sh but in ./configs/textrecog/master/masterv2_ResnetExtra_tableRec_dataset_dynamic_mmfp16.py alphabet_file = '/data_0/pubtabnet/processed_data/recognitionAlphabet.txt' however the processed data didn't exist recognitionAlphabet.txt' and other files that mentioned in xxxxmmfp16.py. Also, files mentioned in ohter py files didn't satisfied to the data_preprocess.py file

Hi. '/data_0/pubtabnet/processed_data/recognitionAlphabet.txt‘ is the path I use in the experiment. The alphabet will generate by 'data_preprocess.py', see here The alphabet path is set by here

myh12138 commented 3 years ago

after run python ./table_recognition/data_preprocess.py then, i want to train it sh ./table_recognition/table_text_line_recognition_dist_train.sh but in ./configs/textrecog/master/masterv2_ResnetExtra_tableRec_dataset_dynamic_mmfp16.py alphabet_file = '/data_0/pubtabnet/processed_data/recognitionAlphabet.txt' however the processed data didn't exist recognitionAlphabet.txt' and other files that mentioned in xxxxmmfp16.py. Also, files mentioned in ohter py files didn't satisfied to the data_preprocess.py file

Hi. '/data_0/pubtabnet/processed_data/recognitionAlphabet.txt‘ is the path I use in the experiment. The alphabet will generate by 'data_preprocess.py', see here The alphabet path is set by here

maybe just the name is different, later you use test_ann_files = {'table_Rec_val_debug_0': '/data_0/pubtabnet/processed_data/recognition_val.txt'} and so on but after i run data_preprocess.py , I only get the structure like the readme.