TTA
/
kr_lp_pgnet
18
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
28 Commits
1 Branch
0 Tags
125 KiB
 
 
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from 'db2ef9b4ec'
${ noResults }
Go to file
songhyeonsoo db2ef9b4ec
generate_synthetic: stratified Korean/region sampling + boost 2-line weight 		4 weeks ago
configs fix: lower lr 0.001→0.0001, warmup 5→15, epoch 100→200; add 하/호/배 to HANGUL_CHAR_MAP 1 month ago
data_gen generate_synthetic: stratified Korean/region sampling + boost 2-line weight 4 weeks ago
dict Add Korean LP dictionary and PGNet config 1 month ago
scripts run_step1: add timestamp to output dir and wandb run name, reset synth data on each run 4 weeks ago
tools fix: tight polygon 라벨 + eval GT mat 생성 파이프라인 추가 1 month ago
.gitignore Add data_gen/setup_assets.sh for synthetic LP assets 1 month ago
HANDOFF.md docs: update HANDOFF with diagnosis history and gap-removal experiment 4 weeks ago
README.md docs: separate README (public) and ONBOARDING (Claude handoff) 4 weeks ago

README.md

kr_lp_pgnet

PaddleOCR PGNet 기반 한국 번호판(LP) end-to-end 검출 + OCR 학습 프로젝트.

대상 번호판: 승용(흰), 영업용(노란), 전기차(파란 8자리), 화물·특수.

디렉토리 구조

kr_lp_pgnet/
├── configs/        # PGNet 학습 config (.yml)
├── dict/           # 문자 사전 (kr_lp_dict.txt)
├── data_gen/       # 합성 LP 이미지 생성기
├── scripts/        # 서버 셋업·학습 실행 셸 스크립트
└── tools/          # 라벨 검증·시각화 등 보조 스크립트

서버 측 실행 순서

# 1. 최초 1회: 환경 셋업
docker exec kr_lp_pgnet bash scripts/setup_server.sh

# 2. Step1: 합성 데이터 생성 + 학습
docker exec -d kr_lp_pgnet bash scripts/run_step1.sh

환경 요구사항

  • NVIDIA GPU (RTX 4090 이상 권장)
  • CUDA 12+
  • Python 3.10
  • PaddlePaddle GPU