Generating Masks and Presence Predictions

Imports

%cd ..
%pwd

%pip install tensorflow

import os
import numpy as np
import matplotlib.pyplot as plt

os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"

plt.rcParams["figure.figsize"] = (20, 10)

from tensorflow.keras.models import load_model
from osgeo import gdal
from PIL import Image
from math import ceil
from typing import Tuple

from insar_eventnet import io, inference

%cd notebooks
%pwd

/home/jovyan/AI-Event-Monitoring/notebooks

'/home/jovyan/AI-Event-Monitoring/notebooks'

Initialize Variables

tile_size = 512
crop_size = 512

mask_model_path = "models/masking_model"
pres_model_path = "models/classification_model"
image_path = input("Image Path: ")
image_name = image_path.split("/")[-1].split(".")[0]
output_path = f"masks_inferred/{image_name}_mask.tif"

mask_model = load_model(mask_model_path)
pres_model = load_model(pres_model_path)

is_wrapped = False

if is_wrapped:
    wrapped, gdal_dataset = io.get_image_array(image_path)
else:
    unwrapped, gdal_dataset = io.get_image_array(image_path)
    wrapped = np.angle(np.exp(1j * (unwrapped)))

Image Path:  SAR_DATA/Positives/S1BA_20210303T162334_20210309T162415_VVR006_INT80_G_ueF_D75C_unw_phase.tif

WARNING:tensorflow:Mixed precision compatibility check (mixed_float16): WARNING
The dtype policy mixed_float16 may run slowly because this machine does not have a GPU. Only Nvidia GPUs with compute capability of at least 7.0 run quickly with mixed_float16.
If you will use compatible GPU(s) not attached to this host, e.g. by running a multi-worker model, you can ignore this warning. This message will only be logged once

/home/jovyan/.local/envs/rtc_analysis/lib/python3.9/site-packages/osgeo/gdal.py:287: FutureWarning: Neither gdal.UseExceptions() nor gdal.DontUseExceptions() has been explicitly called. In GDAL 4.0, exceptions will be enabled by default.
  warnings.warn(

Inference on a Single Wrapped or Unwrapped Image

mask, presence_mask, presence_values = inference.mask_with_model(
    mask_model=mask_model,
    pres_model=pres_model,
    arr_w=wrapped,
    tile_size=tile_size,
    crop_size=crop_size,
)

Plot Results

if np.max(presence_values) > 0.7:
    print("Positive!")
else:
    print("Negative!")

mask[unwrapped == 0] = 0

inference.plot_results(wrapped, mask, presence_mask)

Positive!

Save with Geo-Metadata

if not os.path.exists("masks_inferred"):
    os.makedirs("masks_inferred")

output_img = Image.fromarray(mask)
output_img.save(output_path)

out_dataset = gdal.Open(output_path, gdal.GA_Update)
out_dataset.SetProjection(gdal_dataset.GetProjection())
out_dataset.SetGeoTransform(gdal_dataset.GetGeoTransform())
out_dataset.FlushCache()

0