Datasets:
cassiekang
commited on
Upload cub200_dataset.py
Browse files- cub200_dataset.py +79 -0
cub200_dataset.py
ADDED
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# -*- coding: utf-8 -*-
|
2 |
+
"""cub200_dataset.py
|
3 |
+
|
4 |
+
Automatically generated by Colaboratory.
|
5 |
+
|
6 |
+
Original file is located at
|
7 |
+
https://colab.research.google.com/drive/1qC5RnFLP3_9X50ripGf5YtfXnugxBj2m
|
8 |
+
"""
|
9 |
+
|
10 |
+
from PIL import Image
|
11 |
+
import os
|
12 |
+
import pandas as pd
|
13 |
+
from datasets import DatasetDict, DatasetInfo, Features, Value, Sequence, Image, SplitGenerator, GeneratorBasedBuilder, Version
|
14 |
+
|
15 |
+
_CITATION = """\
|
16 |
+
@techreport{WahCUB_200_2011,
|
17 |
+
Title = {The Caltech-UCSD Birds-200-2011 Dataset},
|
18 |
+
Author = {Wah, C. and Branson, S. and Welinder, P. and Perona, P. and Belongie, S.},
|
19 |
+
Year = {2011},
|
20 |
+
Institution = {California Institute of Technology},
|
21 |
+
Number = {CNS-TR-2011-001}
|
22 |
+
}
|
23 |
+
"""
|
24 |
+
|
25 |
+
_DESCRIPTION = """\
|
26 |
+
The CUB-200-2011 dataset contains 11,788 photos of 200 bird species. Each photo comes with detailed annotations, including part locations, bounding boxes, and attributes for studying fine-grained visual categorization.
|
27 |
+
"""
|
28 |
+
|
29 |
+
_HOMEPAGE = "http://www.vision.caltech.edu/visipedia/CUB-200-2011.html"
|
30 |
+
|
31 |
+
_DATASET_PATH = "/content/drive/My Drive/cub200/CUB_200_2011"
|
32 |
+
|
33 |
+
class CUB2002011(datasets.GeneratorBasedBuilder):
|
34 |
+
"""CUB-200-2011 dataset for bird species image classification."""
|
35 |
+
|
36 |
+
# Version of the dataset
|
37 |
+
VERSION = datasets.Version("1.0.0")
|
38 |
+
|
39 |
+
# Define the features of the dataset, including the image and the label
|
40 |
+
def _info(self):
|
41 |
+
return datasets.DatasetInfo(
|
42 |
+
description="CUB-200-2011 is an image dataset with photos of 200 bird species.",
|
43 |
+
features=datasets.Features({
|
44 |
+
"image": datasets.Image(),
|
45 |
+
"label": datasets.ClassLabel(names=[f"species_{i:03d}" for i in range(1, 201)]),
|
46 |
+
}),
|
47 |
+
supervised_keys=("image", "label"),
|
48 |
+
homepage="http://www.vision.caltech.edu/visipedia/CUB-200-2011.html",
|
49 |
+
citation="""@techreport{WahCUB_200_2011,
|
50 |
+
Title = {The Caltech-UCSD Birds-200-2011 Dataset},
|
51 |
+
Author = {Wah, C. and Branson, S. and Welinder, P. and Perona, P. and Belongie, S.},
|
52 |
+
Year = {2011},
|
53 |
+
Institution = {California Institute of Technology},
|
54 |
+
Number = {CNS-TR-2011-001}
|
55 |
+
}"""
|
56 |
+
)
|
57 |
+
|
58 |
+
# Specify the dataset splits
|
59 |
+
def _split_generators(self, dl_manager):
|
60 |
+
# Assuming the dataset is pre-downloaded
|
61 |
+
dl_manager = DownloadManager.download_and_extract("https://data.caltech.edu/records/65de6-vp158/files/CUB_200_2011.tgz")
|
62 |
+
return [
|
63 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"data_dir": data_dir, "split": "train"}),
|
64 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"data_dir": data_dir, "split": "test"}),
|
65 |
+
]
|
66 |
+
|
67 |
+
# Generate examples from the dataset directory
|
68 |
+
def _generate_examples(self, data_dir, split):
|
69 |
+
# Implement logic to iterate over the dataset and yield examples
|
70 |
+
# For simplicity, assuming all images are in the 'images' folder and split is ignored
|
71 |
+
species_dirs = [p for p in (data_dir / "images").iterdir() if p.is_dir()]
|
72 |
+
for species_dir in species_dirs:
|
73 |
+
species_label = species_dir.name
|
74 |
+
for image_path in species_dir.glob("*.jpg"):
|
75 |
+
# The key can be whatever unique identifier you choose; here we use the image path
|
76 |
+
yield image_path.stem, {
|
77 |
+
"image": str(image_path),
|
78 |
+
"label": species_label,
|
79 |
+
}
|