sbrl-github/research-rainfallradar
1
0
Fork 0
mirror of https://github.com/sbrl/research-rainfallradar synced 2024-07-02 11:24:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

rainfallwrangler json2tfrecord.py: normalise data

Browse source
This commit is contained in:
Starbeamrainbowlabs 2022-09-01 19:03:15 +01:00
parent 3e4128c0a8
commit 9edda1f397
Signed by: sbrl
GPG key ID: 1BE5172E637709C2
1 changed files with 6 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
rainfallwrangler/src/lib/python/json2tfrecord.py
View file

@ -11,6 +11,10 @@ if not os.environ.get("NO_SILENCE"):
silence_tensorflow()
import tensorflow as tf
# The maximum value allowed for the rainfall radar data. Used to normalise the data when converting to .tfrecord files
# TODO: Enter the optimal value for this.
RAINFALL_MAX_NUMBER = 100
def parse_args():
parser = argparse.ArgumentParser(description="Convert a generated .jsonl.gz file to a .tfrecord.gz file")
parser.add_argument("--input", "-i", help="Path to the input file to convert.", required=True)
@ -37,7 +41,8 @@ def convert(filepath_in, filepath_out):
rainfall = tf.constant(obj["rainfallradar"], dtype=tf.float32)
water = tf.constant(obj["waterdepth"], dtype=tf.float32)
# TODO: cast float32 → divide by max_value → clip 0-1 (or -1 to +1? I don't know)
# Normalise the rainfall radar data (the water depth data is already normalised as it's just 0 or 1)
rainfall = tf.clip_by_value(rainfall / RAINFALL_MAX_NUMBER, 0, 1)
###
## 3: Print shape definitions (required when parsing)