task-bake

#!/bin/bash


usage() {
  echo "usage: task-bake [--always] [--config file] [--device DEV] [--batch N] [--compression FMT] name source [source ...]"
  echo
  echo "arguments:"
  echo
  echo "  name              : dataset name prefix"
  echo "  source            : source images folder(s)"
  echo
  echo "bake options:"
  echo
  echo "  --always          : always generate dataset even if exists"
  echo "  --config FILE     : read configuration parameters from file"
  echo "  --device DEV      : tensorflow device"
  echo "  --batch N         : bake batch size"
  echo "  --compression FMT : tfr compression"
}


DATASET=
SOURCES=

ALWAYS=0
DEVICE="/cpu:0"
BATCH_SIZE=1

BAKE_COMPRESSION=none
BAKE_TEST_RATIO=0.2
BAKE_EVAL_RATIO=0.1
BAKE_SIZE=256
BAKE_BLUR_RADIUS=0
BAKE_BLUR_SCALE=1.0
BAKE_CENTER_CROP=1.0
BAKE_SCALES="1.0"
BAKE_FLIPS="none"
BAKE_ROTATIONS="0"
BAKE_CROPS=1
BAKE_BRIGHTNESS=0.0
BAKE_CONTRAST=0.0
BAKE_GAUSSIAN_NOISE=0.0
BAKE_UNIFORM_NOISE=0.0

while [ $# -ge 2 ]
do
  case "$1" in
    --always)
      ALWAYS=1
      shift
      ;;
    --config)
      source $2
      shift 2
      ;;
    --device)
      DEVICE=$2
      shift 2
      ;;
    --batch)
      BATCH_SIZE=$2
      shift 2
      ;;
    --compression)
      BAKE_COMPRESSION=$2
      shift 2
      ;;
    *)
      DATASET=$1
      shift
      SOURCES=$@
      shift $#
      ;;
  esac
done
if [ $# -gt 0 -o "x$DATASET" = "x" -o "x$SOURCES" = "x" ]
then
  usage
  exit 1
fi


DATASET_PREFIX="${DATASET}-${BAKE_SIZE}x${BAKE_SIZE}"

if [ $ALWAYS -eq 0 -a -f "dataset/${DATASET_PREFIX}-train.tfr" -a -f "dataset/${DATASET_PREFIX}-eval.tfr" -a -f "dataset/${DATASET_PREFIX}-test.tfr" ]
then
  echo "DATASET: ${DATASET_PREFIX} already exists, skipping..."
  exit 0
fi


mkdir -p bake

LOG_FILE="bake/$DATASET-$(date +%Y%m%d%H%M%S).log"

log_to_file() {
  while IFS= read -r line
  do
    echo "$line" >> "$LOG_FILE"
    echo "$line"
  done
}


python3 src/run.py bake \
  --device $DEVICE \
  --batch $BATCH_SIZE \
  --compression $BAKE_COMPRESSION \
  --testratio $BAKE_TEST_RATIO \
  --evalratio $BAKE_EVAL_RATIO \
  --size $BAKE_SIZE \
  --blur $BAKE_BLUR_RADIUS \
  --blurscale $BAKE_BLUR_SCALE \
  --centercrop $BAKE_CENTER_CROP \
  --scales $BAKE_SCALES \
  --flips $BAKE_FLIPS \
  --rotations $BAKE_ROTATIONS \
  --crops $BAKE_CROPS \
  --brightness $BAKE_BRIGHTNESS \
  --contrast $BAKE_CONTRAST \
  --gnoise $BAKE_GAUSSIAN_NOISE \
  --unoise $BAKE_UNIFORM_NOISE \
  bake/${DATASET} \
  $SOURCES 2>&1 | log_to_file

RETCODE=$?


if [ $RETCODE -eq 0 ]
then
  if [ -d dataset ]
  then
    echo "DATASET: saving data from bake/${DATASET}* to dataset/..."
    mv bake/${DATASET}* dataset/
  fi
else
  echo "DATASET: cleanup bake/${DATASET}*..."
  rm -f bake/${DATASET}*
fi