ocr-table/docs/scripts/table_model.sh

#!/bin/bash
# shellcheck disable=SC2155

set -eux

readonly CUR_TIME=$(date "+%s")

# SLANet_ch 模型是 PaddleOCR 目前最优的中文表格预训练模型
# https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/doc/doc_ch/table_recognition.md
readonly MODEL_NAME="SLANet_ch"
readonly MODEL_CONF="configs/table/${TABLE_NAME}.yml"

# 推理参数
readonly INFER_IMG_DIR="train_data/table-dataset/conv.v16i/all"

edit_model() {
    vim "${MODEL_CONF}"
}

train_model() {
    python3 tools/train.py -c "${MODEL_CONF}"
}

train_model_distr() {
    python3 \
        -m paddle.distributed.launch \
        --gpus '0,1,2,3,4,5,6,7' \
        tools/train.py -c "${MODEL_CONF}"
}

export_model() {
    python3 tools/export_model.py \
        -c "${MODEL_CONF}" \
        -o Global.pretrained_model="./output/${MODEL_NAME}/best_accuracy" \
        Global.save_inference_dir="./inference/${MODEL_NAME}"
}

infer_model() {
    python3 ppstructure/table/predict_structure.py \
        --table_model_dir=inference/"${MODEL_NAME}" \
        --rec_char_dict_path="./ppocr/utils/ppocr_keys_v1.txt" \
        --table_char_dict_path="./ppocr/utils/dict/table_structure_dict_ch.txt" \
        --image_dir="${INFER_IMG_DIR}" \
        --output="inference_results/${MODEL_NAME}_${CUR_TIME}"
}

main() {
    case "${1}" in
    edit)
        edit_model
        ;;
    train)
        train_model
        ;;
    train_distr)
        train_model_distr
        ;;
    export)
        export_model
        ;;
    infer)
        infer_model
        ;;
    *)
        echo "Invalid option: ${1}"
        exit 1
        ;;
    esac
}

main "$@"