Deprecate GDAL and switch to PIL for reading datasets; Update README

README.md

@@ -18,17 +18,17 @@ The library includes python code for the following:
 </p>
 
 ## Installation
-The library is available as a ```pip``` package. To install the package, run the following commands:
+The library is available as a ```pip``` package. To install the package, run the following command:
 
 ```
-sudo apt-get install python3-pip libgdal-dev gdal-bin -y
 python3 -m pip install sgptools
 ```
 
 Installation from source:
 
 ```
-sudo apt-get install libhdf5-dev python3-pip -y
+git clone https://github.com/itskalvik/sgp-tools.git
+cd sgp-tools/
 python3 -m pip install -r requirements.txt
 python3 -m pip install -e .
 ```
@@ -56,6 +56,7 @@ Please refer to the [examples](https://github.com/itskalvik/sgp-tools/tree/main/
  - `augmented_gpr.py`: GPflow's GP that supports transforms (expansion and aggregation)
  - `augmented_sgpr.py`: GPflow's SGP that supports transforms (expansion and aggregation)
  - `transformations.py`: Expansion and aggregation transforms for IPP
+ - `osgpr.py`: Thang Bui's implementation of online sparse variational GP used for online/adaptive IPP
  - `bo.py`: Bayesian optimization-based sensor placement method that maximizes mutual information
  - `cma_es.py`: Genetic algorithm-based sensor placement method that maximizes mutual information
  - `continuous_sgp.py`: Continuous SGP-based sensor placement method
@@ -69,7 +70,7 @@ Please refer to the [examples](https://github.com/itskalvik/sgp-tools/tree/main/
  - `tsp.py`: TSP solver
 
 ## Datasets
-* The US elevation datasets can be downloaded from [here](https://coast.noaa.gov/digitalcoast/)
+* High-resolution topography and bathymetry datasets can be downloaded from [NOAA Digital Coast](https://coast.noaa.gov/digitalcoast/).
 
 ## About SGP-Tools
 Please consider citing the following papers if you use SGP-Tools in your academic work :smile:

docs/index.md

@@ -13,22 +13,22 @@ The library includes python code for the following:
 <p align="center"><img src="assets/point_sensing.gif"><img src="assets/non-point_sensing.gif"></p>
 
 ## Installation
-The library is available as a ```pip``` package. To install the package, run the following commands:
+The library is available as a ```pip``` package. To install the package, run the following command:
 
 ```
-sudo apt-get install python3-pip libgdal-dev gdal-bin -y
 python3 -m pip install sgptools
 ```
 
 Installation from source:
 
 ```
-sudo apt-get install libhdf5-dev python3-pip -y
+git clone https://github.com/itskalvik/sgp-tools.git
+cd sgp-tools/
 python3 -m pip install -r requirements.txt
 python3 -m pip install -e .
 ```
 
-Note: The requirements.txt file contains packages and their latest versions that were verified to be working without any issues.
+Note: The requirements.txt file contains packages and their latest versions that were last verified to be working without any issues.
 
 ## Quick Start
 Please refer to the [examples](https://github.com/itskalvik/sgp-tools/tree/main/examples) folder for Jupyter notebooks demonstrating all the methods included in the library 😄

requirements.txt

@@ -13,4 +13,4 @@ cma==3.4.0
 bayesian-optimization==1.5.1
 hkb-diamondsquare==1.0.2
 typing_extensions==4.12.2
-GDAL==3.4.1
+pillow==10.4.0

setup.py

@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-__version__ = "1.0.3"
+__version__ = "1.0.4"
 
 setup(
  name='sgptools',
@@ -29,6 +29,7 @@
  'tensorflow>=2.13.0; platform_machine!="arm64"',
  'tensorflow-aarch64>=2.13.0; platform_machine=="arm64"',
  'typing_extensions',
- 'gpflow>=2.7.0'
+ 'gpflow>=2.7.0',
+ 'pillow'
  ]
 )

sgptools/utils/data.py

@@ -18,12 +18,8 @@
 from sklearn.preprocessing import StandardScaler
 from hkb_diamondsquare.DiamondSquare import diamond_square
 
-# Load optional dependency 
-try:
- from osgeo import gdal
-except:
- pass
-
+import PIL
+PIL.Image.MAX_IMAGE_PIXELS = 317500000
 
 ####################################################
 # Utils used to prepare synthetic datasets 
@@ -98,16 +94,23 @@ def prep_tif_dataset(dataset_path):
  X: (n, d); Dataset input features
  y: (n, 1); Dataset labels
  '''
- ds = gdal.Open(dataset_path)
- cols = ds.RasterXSize
- rows = ds.RasterYSize
- band = ds.GetRasterBand(1)
- data = band.ReadAsArray(0, 0, cols, rows)
- data[np.where(data==-999999.0)] = np.nan
-
- x1, x2 = np.where(np.isfinite(data))
- X = np.vstack([x1, x2]).T
- y = data[x1, x2].reshape(-1, 1)
+ data = PIL.Image.open(dataset_path)
+ data = np.array(data)
+
+ # create x and y coordinates from the extent
+ x_coords = np.arange(0, data.shape[1])/10
+ y_coords = np.arange(data.shape[0], 0, -1)/10
+ xx, yy = np.meshgrid(x_coords, y_coords)
+ X = np.c_[xx.ravel(), yy.ravel()]
+ y = data.ravel()
+
+ # Remove invalid labels
+ y[np.where(y==-999999.0)] = np.nan
+ X = X[~np.isnan(y)]
+ y = y[~np.isnan(y)]
+
+ X = X.reshape(-1, 2)
+ y = y.reshape(-1, 1)
 
  return X.astype(float), y.astype(float)