File size: 2,676 Bytes
2ca1e9b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
import datasets
import pandas as pd
from datasets import DownloadManager
class SetClassification(datasets.GeneratorBasedBuilder):
"""Set-Classification Images dataset"""
def __init__(self, data_path, *args, **kwargs):
super(SetClassification, self).__init__(*args, **kwargs)
self.data_path = data_path
self.labels = pd.read_csv(f'{self.data_path}/labels.csv')
self.train = self.labels[self.labels['split'] == 'train']
self.test = self.labels[self.labels['split'] == 'test']
self.dl_manager = DownloadManager()
def _info(self):
return datasets.DatasetInfo(
description='Set Classification Images dataset',
)
def _split_generators(self, dl_manager):
return [
datasets.SplitGenerator(
name=datasets.Split.TRAIN,
gen_kwargs={
'images': [f"{self.data_path}/images/{image.filename}" for image in self.train.itertuples()],
'labels': {
'no': [image.no for image in self.train.itertuples()],
'shape': [image.shape for image in self.train.itertuples()],
'color': [image.color for image in self.train.itertuples()],
'shading': [image.shading for image in self.train.itertuples()]
}
}
),
datasets.SplitGenerator(
name=datasets.Split.TEST,
gen_kwargs={
'images': [f"{self.data_path}/images/{image.filename}" for image in self.test.itertuples()],
'labels': {
'no': [image.no for image in self.test.itertuples()],
'shape': [image.shape for image in self.test.itertuples()],
'color': [image.color for image in self.test.itertuples()],
'shading': [image.shading for image in self.test.itertuples()]
}
}
)
]
def _generate_examples(self, images, labels):
for img, label in zip(images, zip(*labels.values())):
try:
with open(img, 'rb') as img_obj:
no, shape, color, shading = label
yield img, {
'image': {"path": img, "bytes": img_obj.read()},
'no': no,
'shape': shape,
'color': color,
'shading': shading
}
except Exception as e:
print(f"Error processing image {img}: {e}") |