//
Search
🔨

[1_5] MMOCR 에서 제공되는 임의의 recognition 모델을 1% (또는 10%) 물류데이터로 학습한다.

상태
In progress
담당
마감일
2022/11/28
요약
선행 태스크 상태
In progress
후행 태스크
관련 마일스톤
2 more properties
Aihub 물류 test score
from the scratch (Aihub 물류 train)
pretrained (no transfer learned)
transfer learned (Aihub 물류 train)
1% word acc
할필요없음
할필요없음
10% word acc
할필요없음
할필요없음
100% word acc
할필요없음
할필요없음
•
1% transfer learned
nohup tools/dist_train.sh \ configs/textrecog/sar/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained.py \ 2 > nohup.out &
Bash
복사
train
cp work_dirs/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained/epoch_100.pth \ pretrained/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained_stilted-grass-112.pth nohup tools/dist_test.sh \ configs/textrecog/sar/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained.py \ pretrained/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained_stilted-grass-112.pth \ 2 > nohup.out &
Bash
복사
eval
11/29 01:45:16 - mmengine - INFO - Epoch(val) [100][305/305] AihubTransit/recog/word_acc: 0.9467 AihubTransit/recog/word_acc_ignore_case: 0.9477 AihubTransit/recog/word_acc_ignore_case_symbol: 0.9499 IC15/recog/word_acc: 0.9470 IC15/recog/word_acc_ignore_case: 0.9490 IC15/recog/word_acc_ignore_case_symbol: 0.9533 AihubTransit/recog/char_recall: 0.9678 AihubTransit/recog/char_precision: 0.9658 IC15/recog/char_recall: 0.9704 IC15/recog/char_precision: 0.9657
Bash
복사
eval ret
python3 -m mmocr.ocr \ --det-config configs/textdet/dbnet/dbnet_resnet18_fpnc_2e_aihubtransit100of100.py \ --det-ckpt pretrained/dbnet_resnet18_fpnc_2e_aihubtransit100of100_hopeful-leaf-117.pth \ --recog-config configs/textrecog/sar/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained.py \ --recog-ckpt pretrained/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained_stilted-grass-112.pth \ data/det/aihub_transit/part_1of100/train/IMG_OCR_6_T_BL-F001_0020.png \ --img-out-dir work_dirs/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained \ --pred-out-file work_dirs/sar_resnet31_parallel-decoder_100e_aihubtransit1of100_pretrained/output.pkl \ --device cpu
Bash
복사
e2e