|
|
import torch |
|
|
from torch.utils.data import Dataset, DataLoader |
|
|
import os |
|
|
import rasterio |
|
|
import numpy as np |
|
|
from datetime import date |
|
|
from pyproj import Transformer |
|
|
|
|
|
S3_OLCI_SCALE = [0.0139465,0.0133873,0.0121481,0.0115198,0.0100953,0.0123538,0.00879161,0.00876539, |
|
|
0.0095103,0.00773378,0.00675523,0.0071996,0.00749684,0.0086512,0.00526779,0.00530267, |
|
|
0.00493004,0.00549962,0.00502847,0.00326378,0.00324118] |
|
|
|
|
|
Cls_index_binary = { |
|
|
'invalid': 0, |
|
|
'clear': 1, |
|
|
'cloud': 2, |
|
|
} |
|
|
|
|
|
Cls_index_multi = { |
|
|
'invalid': 0, |
|
|
'clear': 1, |
|
|
'cloud-sure': 2, |
|
|
'cloud-ambiguous': 3, |
|
|
'cloud shadow': 4, |
|
|
'snow and ice': 5, |
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
class S3OLCI_CloudDataset(Dataset): |
|
|
''' |
|
|
1596/399 train/test images 256x256 |
|
|
21 bands |
|
|
nodata: nan |
|
|
|
|
|
''' |
|
|
def __init__(self, root_dir, split='train', mode='multi', meta=True): |
|
|
self.root_dir = root_dir |
|
|
self.meta = meta |
|
|
|
|
|
self.img_dir = os.path.join(root_dir, split, 's3_olci') |
|
|
self.fpaths = os.listdir(self.img_dir) |
|
|
self.fpaths = [f for f in self.fpaths if f.endswith('.tif')] |
|
|
|
|
|
if mode == 'multi': |
|
|
self.cloud_dir = os.path.join(root_dir, split, 'cloud_multi') |
|
|
elif mode == 'binary': |
|
|
self.cloud_dir = os.path.join(root_dir, split, 'cloud_binary') |
|
|
|
|
|
if self.meta: |
|
|
self.reference_date = date(1970, 1, 1) |
|
|
|
|
|
|
|
|
def __len__(self): |
|
|
return len(self.fpaths) |
|
|
|
|
|
def __getitem__(self, idx): |
|
|
fpath = self.fpaths[idx] |
|
|
fpath_img = os.path.join(self.img_dir, fpath) |
|
|
fpath_cloud = os.path.join(self.cloud_dir, fpath) |
|
|
|
|
|
with rasterio.open(fpath_img) as src: |
|
|
img = src.read() |
|
|
|
|
|
img[np.isnan(img)] = 0 |
|
|
|
|
|
for b in range(21): |
|
|
img[b] = img[b] * S3_OLCI_SCALE[b] |
|
|
|
|
|
if self.meta: |
|
|
cx,cy = src.xy(src.height // 2, src.width // 2) |
|
|
crs_transformer = Transformer.from_crs(src.crs, 'epsg:4326') |
|
|
lon, lat = crs_transformer.transform(cx,cy) |
|
|
img_fname = os.path.basename(fpath_img) |
|
|
date_str = img_fname.split('____')[1][:8] |
|
|
date_obj = date(int(date_str[:4]), int(date_str[4:6]), int(date_str[6:8])) |
|
|
delta = (date_obj - self.reference_date).days |
|
|
meta_info = np.array([lon, lat, delta, np.nan]).astype(np.float32) |
|
|
else: |
|
|
meta_info = np.array([np.nan,np.nan,np.nan,np.nan]).astype(np.float32) |
|
|
|
|
|
img = torch.from_numpy(img).float() |
|
|
|
|
|
with rasterio.open(fpath_cloud) as src: |
|
|
cloud = src.read(1) |
|
|
cloud = torch.from_numpy(cloud).long() |
|
|
|
|
|
return img, cloud, meta_info |
|
|
|
|
|
|
|
|
|
|
|
if __name__ == '__main__': |
|
|
dataset = S3OLCI_CloudDataset(root_dir='./cloud_s3olci', split='train', mode='multi') |
|
|
dataloader = DataLoader(dataset, batch_size=2, shuffle=False) |
|
|
for img, cloud, meta in dataloader: |
|
|
print(img.shape, cloud.shape, meta.shape) |
|
|
break |
