Cheatsheets

# Variables
tile_size = 256
class_name = "building"

# Lists
class_names = ["background", "building", "road"]
first_class = class_names[0]

# Dictionaries
metrics = {"iou": 0.82, "f1": 0.90}
print(metrics["iou"])

# Loops
for name in class_names:
    print(name)

# Functions
def to_hectares(area_m2):
    return area_m2 / 10_000

# Imports
import geoai
import numpy as np
from pathlib import Path

# Paths
data_path = Path("data") / "raw" / "image.tif"

import pandas as pd
import geopandas as gpd

# Reading data
df = pd.read_csv("results.csv")
gdf = gpd.read_file("buildings.geojson")

# Inspecting columns
print(df.columns)
print(gdf.head())

# Filtering rows
large_buildings = gdf[gdf["area_m2"] > 100]

# Selecting columns
subset = gdf[["geometry", "area_m2", "class"]]

# Plotting a simple output
gdf.plot(column="area_m2", legend=True)

# Saving results
gdf.to_file("filtered_buildings.geojson", driver="GeoJSON")
df.to_csv("summary.csv", index=False)

import rasterio as rio
import numpy as np

# Opening raster data
with rio.open("image.tif") as src:
    # Inspecting metadata
    print(src.count, src.width, src.height, src.crs, src.res)

    # Reading bands
    band_1 = src.read(1)
    rgb = src.read([1, 2, 3])

    # Handling nodata values
    nodata_value = src.nodata
    valid = band_1 != nodata_value

# Plotting an array
import matplotlib.pyplot as plt
plt.imshow(band_1, cmap="viridis")
plt.colorbar()
plt.show()

# Saving outputs (matching the input's profile)
with rio.open("image.tif") as src:
    profile = src.profile

with rio.open("output.tif", "w", **profile) as dst:
    dst.write(band_1, 1)

from pathlib import Path

project_root = Path.cwd()
data_dir = project_root / "data"
output_dir = project_root / "outputs"

output_dir.mkdir(exist_ok=True)

# Build a path without worrying about slashes
image_path = data_dir / "raw" / "scene_2024.tif"

fig, ax = plt.subplots(figsize=(8, 6))
ax.imshow(band_1, cmap="viridis")

ax.set_title("Study Area NDVI")   # Title
ax.set_xlabel("Column")            # Labels
ax.set_ylabel("Row")

# Legends (for categorical or vector plots)
gdf.plot(ax=ax, column="class", legend=True)

# Saving a figure
fig.savefig("outputs/ndvi_map.png", dpi=300, bbox_inches="tight")

git status
git add .
git commit -m "Add NDVI tiling script"
git push