geirhos_conflict_stimuli
Stay organized with collections
Save and categorize content based on your preferences.
Shape/texture conflict stimuli from "ImageNet-trained CNNs are biased towards
texture; increasing shape bias improves accuracy and robustness."
Note that, although the dataset source contains images with matching shape and
texture and we include them here, they are ignored for most evaluations in the
original paper.
Split |
Examples |
'test' |
1,280 |
FeaturesDict({
'file_name': Text(shape=(), dtype=string),
'image': Image(shape=(None, None, 3), dtype=uint8),
'shape_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1000)),
'shape_label': ClassLabel(shape=(), dtype=int64, num_classes=16),
'texture_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1000)),
'texture_label': ClassLabel(shape=(), dtype=int64, num_classes=16),
})
Feature |
Class |
Shape |
Dtype |
Description |
|
FeaturesDict |
|
|
|
file_name |
Text |
|
string |
|
image |
Image |
(None, None, 3) |
uint8 |
|
shape_imagenet_labels |
Sequence(ClassLabel) |
(None,) |
int64 |
|
shape_label |
ClassLabel |
|
int64 |
|
texture_imagenet_labels |
Sequence(ClassLabel) |
(None,) |
int64 |
|
texture_label |
ClassLabel |
|
int64 |
|

@inproceedings{
geirhos2018imagenettrained,
title={ImageNet-trained {CNN}s are biased towards texture; increasing shape
bias improves accuracy and robustness.},
author={Robert Geirhos and Patricia Rubisch and Claudio Michaelis and
Matthias Bethge and Felix A. Wichmann and Wieland Brendel},
booktitle={International Conference on Learning Representations},
year={2019},
url={https://openreview.net/forum?id=Bygh9j09KX},
}
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2022-12-06 UTC.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2022-12-06 UTC."],[],[],null,["# geirhos_conflict_stimuli\n\n\u003cbr /\u003e\n\n- **Description**:\n\nShape/texture conflict stimuli from \"ImageNet-trained CNNs are biased towards\ntexture; increasing shape bias improves accuracy and robustness.\"\n\nNote that, although the dataset source contains images with matching shape and\ntexture and we include them here, they are ignored for most evaluations in the\noriginal paper.\n\n- **Homepage** :\n \u003chttps://github.com/rgeirhos/texture-vs-shape\u003e\n\n- **Source code** :\n [`tfds.image_classification.GeirhosConflictStimuli`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/image_classification/geirhos_conflict_stimuli.py)\n\n- **Versions**:\n\n - **`1.0.0`** (default): No release notes.\n- **Download size** : `153.96 MiB`\n\n- **Dataset size** : `130.44 MiB`\n\n- **Auto-cached**\n ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):\n Only when `shuffle_files=False` (test)\n\n- **Splits**:\n\n| Split | Examples |\n|----------|----------|\n| `'test'` | 1,280 |\n\n- **Feature structure**:\n\n FeaturesDict({\n 'file_name': Text(shape=(), dtype=string),\n 'image': Image(shape=(None, None, 3), dtype=uint8),\n 'shape_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1000)),\n 'shape_label': ClassLabel(shape=(), dtype=int64, num_classes=16),\n 'texture_imagenet_labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=1000)),\n 'texture_label': ClassLabel(shape=(), dtype=int64, num_classes=16),\n })\n\n- **Feature documentation**:\n\n| Feature | Class | Shape | Dtype | Description |\n|-------------------------|----------------------|-----------------|--------|-------------|\n| | FeaturesDict | | | |\n| file_name | Text | | string | |\n| image | Image | (None, None, 3) | uint8 | |\n| shape_imagenet_labels | Sequence(ClassLabel) | (None,) | int64 | |\n| shape_label | ClassLabel | | int64 | |\n| texture_imagenet_labels | Sequence(ClassLabel) | (None,) | int64 | |\n| texture_label | ClassLabel | | int64 | |\n\n- **Supervised keys** (See\n [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):\n `('image', 'shape_label')`\n\n- **Figure**\n ([tfds.show_examples](https://www.tensorflow.org/datasets/api_docs/python/tfds/visualization/show_examples)):\n\n- **Examples** ([tfds.as_dataframe](https://www.tensorflow.org/datasets/api_docs/python/tfds/as_dataframe)):\n\nDisplay examples... \n\n- **Citation**:\n\n @inproceedings{\n geirhos2018imagenettrained,\n title={ImageNet-trained {CNN}s are biased towards texture; increasing shape\n bias improves accuracy and robustness.},\n author={Robert Geirhos and Patricia Rubisch and Claudio Michaelis and\n Matthias Bethge and Felix A. Wichmann and Wieland Brendel},\n booktitle={International Conference on Learning Representations},\n year={2019},\n url={https://openreview.net/forum?id=Bygh9j09KX},\n }"]]