legacytraining.sh

#!/bin/bash

FONT="LEGO Hatting AM"
FONTS_DIR="/home/dk8prina/fonts"
TRAINING_DIR="/mnt/c/Users/dk8PriNa/LEGO/Image-test/src/output/synthetic_samplesv1v2Ver2"
TEXT_DIR="/mnt/c/Users/dk8PriNa/LEGO/Image-test/src/output/synthetic_samplesv1v2Ver2/training_text.txt"
TESSDATA="/usr/share/tesseract-ocr/5/tessdata"
OUTPUT_MODEL="Hattingam_legacyV1V2_Ver2"
RESOLUTION=300
NUM_IMAGES=100

cd "$TRAINING_DIR" || exit 1

# for ((i=50; i<NUM_IMAGES; i++)); do
#         text2image \
#         --text "$TEXT_DIR" \
#         --outputbase "$TRAINING_DIR/A_${i}" \
#         --font="$FONT" \
#         --fonts_dir="$FONTS_DIR" \
#         --ptsize $((20 + RANDOM % 20)) \
#         --resolution $RESOLUTION \
#         --fontconfig_tmpdir=/tmp

#         cp "$TEXT_DIR" "$TRAINING_DIR/A_${i}.gt.txt"
# done

find "$TRAINING_DIR" -type f -name "*.tr" -size 0c -delete

echo "Generating .box and .tr files..."
find "$TRAINING_DIR" -type f -name "*.tif" | while read -r img; do
    base=$(basename "$img" .tif)
    dir=$(dirname "$img")
    gt_file="$dir/$base.gt.txt"

    if [[ -f "$gt_file" && -s "$img" ]]; then
        echo "Processing $img"
        tesseract "$img" "$dir/$base" --psm 6 box.train
    else
        echo  "Skipping $img (missing $gt_file or empty image)"
    fi
done

BOX_FILES=$(find "$TRAINING_DIR" -type f -name "*.box" -size +0c)
TR_FILES=$(find "$TRAINING_DIR" -type f -name "*.tr" -size +0c)


if [[ -z "$BOX_FILES" ]]; then
    echo "Error: No valid .box files found! Training aborted."
    exit 1
fi

echo "Getting unicharset"
unicharset_extractor $BOX_FILES

echo "Getting the font properties"
echo "$OUTPUT_MODEL 0 0 0 0 0" > font_properties

echo "Creating the shapeclustering file"
shapeclustering -F font_properties -U unicharset $TR_FILES

echo "Running mftraining"
mftraining -F font_properties -U unicharset -O unicharset $TR_FILES


if [[ -z "$TR_FILES" ]]; then
    echo "Error: No valid .tr files found! Training aborted."
    exit 1
fi


for tr in $TR_FILES; do
    if ! grep -q "TrainingPage" "$tr"; then
        echo "Invalid TR file: $tr"
    fi
done

echo "Running cntraining"
cntraining $TR_FILES

echo "Renaming output files..."
mv inttemp "$OUTPUT_MODEL.inttemp"
mv normproto "$OUTPUT_MODEL.normproto"
mv pffmtable "$OUTPUT_MODEL.pffmtable"
mv shapetable "$OUTPUT_MODEL.shapetable"
mv unicharset "$OUTPUT_MODEL.unicharset"

echo "Combining into .traineddata..."
combine_tessdata "$OUTPUT_MODEL."

echo "✅ Legacy training completed! Model saved as $OUTPUT_MODEL.traineddata"