keras-climate 0.1.0__tar.gz

keras_climate-0.1.0/PKG-INFO

@@ -0,0 +1,49 @@
+Metadata-Version: 2.4
+Name: keras_climate
+Version: 0.1.0
+Summary: Keras 3 framework for climate modeling and Earth observation
+Project-URL: Homepage, https://github.com/anas-rz/keras-climate
+Project-URL: Documentation, https://anas-rz.github.io/keras-climate/
+Project-URL: Repository, https://github.com/anas-rz/keras-climate
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Requires-Dist: keras>=3.0
+Requires-Dist: numpy
+
+# keras_climate
+
+![Keras Climate](docs/banner.png)
+
+A Keras 3 (TensorFlow / JAX / PyTorch backend) framework of models for
+climate modeling and Earth observation, plus tooling to port pretrained
+weights from the original reference implementations (almost always
+PyTorch) onto the Keras equivalents.
+
+**[Full documentation](https://anas-rz.github.io/keras-climate/)**
+
+## Install
+
+```bash
+pip install git+https://github.com/anas-rz/keras-climate
+```
+
+## Quick start
+
+```python
+from keras_climate.remote_sensing import UNet, SegFormer
+from keras_climate.forecasting import PatchTST
+from keras_climate.weather import ConvLSTMNowcaster
+from keras_climate.foundation import PrithviClassifier
+
+model = UNet(input_shape=(256, 256, 4), num_classes=5)   # 4-band (e.g. RGB+NIR) input
+model.summary()
+
+forecaster = PatchTST(seq_len=336, pred_len=96, num_channels=7)
+```
+
+All builders return a plain `keras.Model`, so standard `model.compile(...)`
+/ `model.fit(...)` / `model.save(...)` all work unchanged.
+
+See the [documentation](https://anas-rz.github.io/keras-climate/) for the
+full model roster, the pretrained-weight-porting guide, and which models
+ship a ready-to-download real checkpoint.

keras_climate-0.1.0/README.md

@@ -0,0 +1,37 @@
+# keras_climate
+
+![Keras Climate](docs/banner.png)
+
+A Keras 3 (TensorFlow / JAX / PyTorch backend) framework of models for
+climate modeling and Earth observation, plus tooling to port pretrained
+weights from the original reference implementations (almost always
+PyTorch) onto the Keras equivalents.
+
+**[Full documentation](https://anas-rz.github.io/keras-climate/)**
+
+## Install
+
+```bash
+pip install git+https://github.com/anas-rz/keras-climate
+```
+
+## Quick start
+
+```python
+from keras_climate.remote_sensing import UNet, SegFormer
+from keras_climate.forecasting import PatchTST
+from keras_climate.weather import ConvLSTMNowcaster
+from keras_climate.foundation import PrithviClassifier
+
+model = UNet(input_shape=(256, 256, 4), num_classes=5)   # 4-band (e.g. RGB+NIR) input
+model.summary()
+
+forecaster = PatchTST(seq_len=336, pred_len=96, num_channels=7)
+```
+
+All builders return a plain `keras.Model`, so standard `model.compile(...)`
+/ `model.fit(...)` / `model.save(...)` all work unchanged.
+
+See the [documentation](https://anas-rz.github.io/keras-climate/) for the
+full model roster, the pretrained-weight-porting guide, and which models
+ship a ready-to-download real checkpoint.

keras_climate-0.1.0/keras_climate/__init__.py

@@ -0,0 +1,16 @@
+__version__ = "0.1.0"
+
+from keras_climate import remote_sensing
+from keras_climate import forecasting
+from keras_climate import weather
+from keras_climate import foundation
+from keras_climate import weights
+
+__all__ = [
+    "remote_sensing",
+    "forecasting",
+    "weather",
+    "foundation",
+    "weights",
+    "__version__",
+]

keras_climate-0.1.0/keras_climate/datasets/__init__.py

@@ -0,0 +1,405 @@
+"""Multi-backend Keras port of torchgeo.datasets.
+
+Core framework: :class:`GeoDataset`, :class:`RasterDataset`,
+:class:`VectorDataset`, :class:`XarrayDataset`, :class:`NonGeoDataset`,
+:class:`NonGeoClassificationDataset`, :class:`IntersectionDataset`, and
+:class:`UnionDataset`, plus dataset splitting utilities and collation
+helpers, plus the full catalog of concrete leaf datasets ported from
+torchgeo (EuroSAT, So2Sat, Sentinel, SpaceNet, Copernicus-Bench, ...).
+
+Image/mask samples use the channels-last layout (``H x W x C``), a
+deliberate deviation from torchgeo's channels-first convention to match the
+rest of keras_climate. See :mod:`keras_climate.datasets.geo` for details.
+"""
+
+from .copernicus import (
+    CopernicusBenchAQNO2S5P,
+    CopernicusBenchAQO3S5P,
+    CopernicusBenchBase,
+    CopernicusBenchBigEarthNetS1,
+    CopernicusBenchBigEarthNetS2,
+    CopernicusBenchBiomassS3,
+    CopernicusBenchCloudS2,
+    CopernicusBenchCloudS3,
+    CopernicusBenchDFC2020S1,
+    CopernicusBenchDFC2020S2,
+    CopernicusBenchEuroSATS1,
+    CopernicusBenchEuroSATS2,
+    CopernicusBenchFloodS1,
+    CopernicusBenchLC100ClsS3,
+    CopernicusBenchLC100SegS3,
+    CopernicusBenchLCZS2,
+    CopernicusEmbed,
+    CopernicusPretrain,
+)
+from .errors import DatasetNotFoundError, DependencyNotFoundError, RGBBandsMissingError
+from .eurosat import EuroSAT, EuroSAT100, EuroSATSpatial
+from .geo import (
+    GeoDataset,
+    IntersectionDataset,
+    NonGeoClassificationDataset,
+    NonGeoDataset,
+    RasterDataset,
+    UnionDataset,
+    VectorDataset,
+    XarrayDataset,
+)
+from .mixins import PlottingMixin
+from .spacenet import (
+    SpaceNet,
+    SpaceNet1,
+    SpaceNet2,
+    SpaceNet3,
+    SpaceNet4,
+    SpaceNet5,
+    SpaceNet6,
+    SpaceNet7,
+    SpaceNet8,
+)
+from .splits import (
+    random_bbox_assignment,
+    random_bbox_splitting,
+    random_grid_cell_assignment,
+    roi_split,
+    time_series_split,
+)
+from .utils import (
+    BoundingBox,
+    concat_samples,
+    merge_samples,
+    stack_samples,
+    unbind_samples,
+)
+
+from .advance import ADVANCE
+from .agb_live_woody_density import AbovegroundLiveWoodyBiomassDensity
+from .agrifieldnet import AgriFieldNetImage, AgriFieldNetMask, AgriFieldNet
+from .air_quality import AirQuality
+from .airphen import Airphen
+from .astergdem import AsterGDEM
+from .benin_cashews import BeninSmallHolderCashews
+from .bigearthnet import BigEarthNet, BigEarthNetV2
+from .biomassters import BioMassters
+from .bright import BRIGHTDFC2025
+from .cabuar import CaBuAr
+from .caffe import CaFFe
+from .cbf import CanadianBuildingFootprints
+from .cdl import CDL
+from .chabud import ChaBuD
+from .chesapeake import Chesapeake, ChesapeakeDC, ChesapeakeDE, ChesapeakeMD, ChesapeakeNY, ChesapeakePA, ChesapeakeVA, ChesapeakeWV, ChesapeakeCVPR
+from .clay import ClayEmbeddings
+from .cloud_cover import CloudCoverDetection
+from .cms_mangrove_canopy import CMSGlobalMangroveCanopy
+from .cowc import COWC, COWCCounting, COWCDetection
+from .cropharvest import CropHarvest
+from .cv4a_kenya_crop_type import CV4AKenyaCropType
+from .cyclone import TropicalCyclone
+from .deepglobelandcover import DeepGlobeLandCover
+from .dfc2022 import DFC2022
+from .digital_typhoon import DigitalTyphoon
+from .dior import DIOR
+from .dl4gam import DL4GAMAlps
+from .dlrsd import DLRSDBase, DLRSD, DLRSDMultilabel
+from .dota import DOTA
+from .earth_embeddings import EarthEmbeddings
+from .earth_index import EarthIndexEmbeddings
+from .eddmaps import EDDMapS
+from .enmap import EnMAP
+from .enviroatlas import EnviroAtlas
+from .esd import ESDQuantizer, EmbeddedSeamlessData
+from .esri2020 import Esri2020
+from .etci2021 import ETCI2021
+from .eudem import EUDEM
+from .eurocrops import EuroCrops
+from .everwatch import EverWatch
+from .fair1m import FAIR1M
+from .fire_risk import FireRisk
+from .flair import FLAIRHUBBase, FLAIRHUB, FLAIRHUBToy
+from .forestdamage import ForestDamage
+from .ftw import FieldsOfTheWorld
+from .gbif import GBIF
+from .gbm import GlobalBuildingMap
+from .geonrw import GeoNRW
+from .gid15 import GID15
+from .globalmangrovewatch import GlobalMangroveWatch
+from .globbiomass import GlobBiomass
+from .gse import GoogleSatelliteEmbedding
+from .hyspecnet import HySpecNet11k
+from .idtrees import IDTReeS
+from .inaturalist import INaturalist
+from .inria import InriaAerialImageLabeling
+from .iobench import IOBench
+from .l7irish import L7IrishImage, L7IrishMask, L7Irish
+from .l8biome import L8BiomeImage, L8BiomeMask, L8Biome
+from .landcoverai import LandCoverAIBase, LandCoverAIGeo, LandCoverAI, LandCoverAI100
+from .landsat import Landsat, Landsat1, Landsat2, Landsat3, Landsat4MSS, Landsat4TM, Landsat5MSS, Landsat5TM, Landsat7, Landsat8, Landsat9
+from .levircd import LEVIRCDBase, LEVIRCD, LEVIRCDPlus
+from .loveda import LoveDA
+from .major_tom import MajorTOMEmbeddings
+from .mapinwild import MapInWild
+from .mdas import MDAS
+from .meta_chm import MetaCHM
+from .millionaid import MillionAID
+from .mmearth import MMEarth
+from .mmflood import MMFloodComponent, MMFlood
+from .naip import NAIP
+from .nasa_marine_debris import NASAMarineDebris
+from .nccm import NCCM
+from .nlcd import NLCD
+from .openaerialmap import TileUtils, OpenAerialMap
+from .openbuildings import OpenBuildings
+from .openstreetmap import OpenStreetMap
+from .oscd import OSCD, OSCD100
+from .pastis import PASTIS, PASTIS100
+from .patternnet import PatternNet
+from .potsdam import Potsdam2D
+from .presto import PrestoEmbeddings
+from .prisma import PRISMA
+from .quakeset import QuakeSet
+from .reforestree import ReforesTree
+from .resisc45 import RESISC45
+from .rwanda_field_boundary import RwandaFieldBoundary
+from .s2_100k import S2100k
+from .satlas import SatlasPretrain
+from .seasonet import SeasoNet
+from .seco import SeasonalContrastS2
+from .sen12ms import SEN12MS
+from .sentinel import Sentinel, Sentinel1, Sentinel2
+from .skippd import SKIPPD
+from .skyscript import SkyScript
+from .so2sat import So2Sat
+from .soda import SODAA
+from .solar_plants_brazil import SolarPlantsBrazil
+from .south_africa_crop_type import SouthAfricaCropType
+from .south_america_soybean import SouthAmericaSoybean
+from .ssl4eo import SSL4EO, SSL4EOL, SSL4EOS12
+from .ssl4eo_benchmark import SSL4EOLBenchmark
+from .substation import Substation
+from .sustainbench_crop_yield import SustainBenchCropYield
+from .tessera import TesseraEmbeddings
+from .treesatai import TreeSatAI
+from .ucmerced import UCMerced
+from .usavars import USAVars
+from .vaihingen import Vaihingen2D
+from .vhr10 import VHR10
+from .western_usa_live_fuel_moisture import WesternUSALiveFuelMoisture
+from .worldstrat import WorldStrat
+from .xbd import xBD, XView2, xBDDistShift
+from .zuericrop import ZueriCrop
+
+__all__ = (
+    "ADVANCE",
+    "AbovegroundLiveWoodyBiomassDensity",
+    "AgriFieldNet",
+    "AgriFieldNetImage",
+    "AgriFieldNetMask",
+    "AirQuality",
+    "Airphen",
+    "AsterGDEM",
+    "BRIGHTDFC2025",
+    "BeninSmallHolderCashews",
+    "BigEarthNet",
+    "BigEarthNetV2",
+    "BioMassters",
+    "BoundingBox",
+    "CDL",
+    "CMSGlobalMangroveCanopy",
+    "COWC",
+    "COWCCounting",
+    "COWCDetection",
+    "CV4AKenyaCropType",
+    "CaBuAr",
+    "CaFFe",
+    "CanadianBuildingFootprints",
+    "ChaBuD",
+    "Chesapeake",
+    "ChesapeakeCVPR",
+    "ChesapeakeDC",
+    "ChesapeakeDE",
+    "ChesapeakeMD",
+    "ChesapeakeNY",
+    "ChesapeakePA",
+    "ChesapeakeVA",
+    "ChesapeakeWV",
+    "ClayEmbeddings",
+    "CloudCoverDetection",
+    "CopernicusBenchAQNO2S5P",
+    "CopernicusBenchAQO3S5P",
+    "CopernicusBenchBase",
+    "CopernicusBenchBigEarthNetS1",
+    "CopernicusBenchBigEarthNetS2",
+    "CopernicusBenchBiomassS3",
+    "CopernicusBenchCloudS2",
+    "CopernicusBenchCloudS3",
+    "CopernicusBenchDFC2020S1",
+    "CopernicusBenchDFC2020S2",
+    "CopernicusBenchEuroSATS1",
+    "CopernicusBenchEuroSATS2",
+    "CopernicusBenchFloodS1",
+    "CopernicusBenchLC100ClsS3",
+    "CopernicusBenchLC100SegS3",
+    "CopernicusBenchLCZS2",
+    "CopernicusEmbed",
+    "CopernicusPretrain",
+    "CropHarvest",
+    "DFC2022",
+    "DIOR",
+    "DL4GAMAlps",
+    "DLRSD",
+    "DLRSDBase",
+    "DLRSDMultilabel",
+    "DOTA",
+    "DatasetNotFoundError",
+    "DeepGlobeLandCover",
+    "DependencyNotFoundError",
+    "DigitalTyphoon",
+    "EDDMapS",
+    "ESDQuantizer",
+    "ETCI2021",
+    "EUDEM",
+    "EarthEmbeddings",
+    "EarthIndexEmbeddings",
+    "EmbeddedSeamlessData",
+    "EnMAP",
+    "EnviroAtlas",
+    "Esri2020",
+    "EuroCrops",
+    "EuroSAT",
+    "EuroSAT100",
+    "EuroSATSpatial",
+    "EverWatch",
+    "FAIR1M",
+    "FLAIRHUB",
+    "FLAIRHUBBase",
+    "FLAIRHUBToy",
+    "FieldsOfTheWorld",
+    "FireRisk",
+    "ForestDamage",
+    "GBIF",
+    "GID15",
+    "GeoDataset",
+    "GeoNRW",
+    "GlobBiomass",
+    "GlobalBuildingMap",
+    "GlobalMangroveWatch",
+    "GoogleSatelliteEmbedding",
+    "HySpecNet11k",
+    "IDTReeS",
+    "INaturalist",
+    "IOBench",
+    "InriaAerialImageLabeling",
+    "IntersectionDataset",
+    "L7Irish",
+    "L7IrishImage",
+    "L7IrishMask",
+    "L8Biome",
+    "L8BiomeImage",
+    "L8BiomeMask",
+    "LEVIRCD",
+    "LEVIRCDBase",
+    "LEVIRCDPlus",
+    "LandCoverAI",
+    "LandCoverAI100",
+    "LandCoverAIBase",
+    "LandCoverAIGeo",
+    "Landsat",
+    "Landsat1",
+    "Landsat2",
+    "Landsat3",
+    "Landsat4MSS",
+    "Landsat4TM",
+    "Landsat5MSS",
+    "Landsat5TM",
+    "Landsat7",
+    "Landsat8",
+    "Landsat9",
+    "LoveDA",
+    "MDAS",
+    "MMEarth",
+    "MMFlood",
+    "MMFloodComponent",
+    "MajorTOMEmbeddings",
+    "MapInWild",
+    "MetaCHM",
+    "MillionAID",
+    "NAIP",
+    "NASAMarineDebris",
+    "NCCM",
+    "NLCD",
+    "NonGeoClassificationDataset",
+    "NonGeoDataset",
+    "OSCD",
+    "OSCD100",
+    "OpenAerialMap",
+    "OpenBuildings",
+    "OpenStreetMap",
+    "PASTIS",
+    "PASTIS100",
+    "PRISMA",
+    "PatternNet",
+    "PlottingMixin",
+    "Potsdam2D",
+    "PrestoEmbeddings",
+    "QuakeSet",
+    "RESISC45",
+    "RGBBandsMissingError",
+    "RasterDataset",
+    "ReforesTree",
+    "RwandaFieldBoundary",
+    "S2100k",
+    "SEN12MS",
+    "SKIPPD",
+    "SODAA",
+    "SSL4EO",
+    "SSL4EOL",
+    "SSL4EOLBenchmark",
+    "SSL4EOS12",
+    "SatlasPretrain",
+    "SeasoNet",
+    "SeasonalContrastS2",
+    "Sentinel",
+    "Sentinel1",
+    "Sentinel2",
+    "SkyScript",
+    "So2Sat",
+    "SolarPlantsBrazil",
+    "SouthAfricaCropType",
+    "SouthAmericaSoybean",
+    "SpaceNet",
+    "SpaceNet1",
+    "SpaceNet2",
+    "SpaceNet3",
+    "SpaceNet4",
+    "SpaceNet5",
+    "SpaceNet6",
+    "SpaceNet7",
+    "SpaceNet8",
+    "Substation",
+    "SustainBenchCropYield",
+    "TesseraEmbeddings",
+    "TileUtils",
+    "TreeSatAI",
+    "TropicalCyclone",
+    "UCMerced",
+    "USAVars",
+    "UnionDataset",
+    "VHR10",
+    "Vaihingen2D",
+    "VectorDataset",
+    "WesternUSALiveFuelMoisture",
+    "WorldStrat",
+    "XView2",
+    "XarrayDataset",
+    "ZueriCrop",
+    "concat_samples",
+    "merge_samples",
+    "random_bbox_assignment",
+    "random_bbox_splitting",
+    "random_grid_cell_assignment",
+    "roi_split",
+    "stack_samples",
+    "time_series_split",
+    "unbind_samples",
+    "xBD",
+    "xBDDistShift",
+)

keras_climate-0.1.0/keras_climate/datasets/_test_helpers.py

@@ -0,0 +1,33 @@
+"""Shared test-only helpers for keras_climate.datasets tests (not a test module itself)."""
+
+
+def assert_dtype(tensor, expected, jax_fallback=None):
+    """Assert a tensor's dtype equals *expected*, portable across backends.
+
+    Keras backends disagree on whether the raw ``tensor.dtype`` object
+    supports string equality: it works for TensorFlow and JAX, but
+    ``torch.dtype.__eq__`` does not accept a plain string (``torch.float32
+    == "float32"`` is ``False``). ``keras.backend.standardize_dtype``
+    normalizes it into a plain, comparable string on every backend.
+
+    Args:
+        tensor: a Keras tensor.
+        expected: the expected dtype string (e.g. ``"float32"``).
+        jax_fallback: an additional dtype string to also accept, but only
+            on the jax backend. Use this for int64 assertions, since JAX
+            disables 64-bit precision by default (a well-known JAX
+            limitation, not something keras_climate's datasets control),
+            silently downcasting int64 requests to int32.
+    """
+    import keras
+
+    actual = keras.backend.standardize_dtype(tensor.dtype)
+    if jax_fallback is not None and keras.backend.backend() == "jax":
+        assert actual in (expected, jax_fallback), actual
+    else:
+        assert actual == expected, actual
+
+
+def assert_int64_dtype(tensor):
+    """Assert an integer tensor is int64, tolerating JAX's 32-bit default."""
+    assert_dtype(tensor, "int64", jax_fallback="int32")