ntucllab/aclmicroimagenet10

Dataset Card for ACLMicroImageNet10

This Complementary labeled MicroImageNet10 dataset contains 3 human-annotated complementary labels for all 5000 images in the training split of TinyImageNet200.

For more details, please visit our github or paper.

Dataset Structure

Data Instances

A sample from the training set is provided below:

{
	'images': <PIL.PngImagePlugin.PngImageFile image mode=RGB size=64x64 at 0x77CB6DF65F30>, 
	'ord_labels': 0, 
	'cl_labels': [8, 3, 6]
}

Data Fields

• —images: A PIL.Image.Image object containing the 64x64 image.

• —ord_labels: The ordinary labels of the images, and they are labeled from 0 to 9 as follows:

0: sulphur-butterfly 1: backpack 2: cardigan 3: kimono 4: magnetic-compass 5: oboe 6: scandal 7: torch 8: pizza 9: alp

• —cl_labels: Three complementary labels for each image.

Citing

If you find this dataset useful, please cite the following:

@InProceedings{aclimage2025,
  author="Mai, Tan-Ha and Ye, Nai-Xuan and Kuan, Yu-Wei and Lu, Po-Yi and Lin, Hsuan-Tien",
  title="The Unexplored Potential of Vision-Language Models for Generating Large-Scale Complementary-Label Learning Data",
  booktitle="Pacific-Asia Conference on Knowledge Discovery and Data Mining",
  year="2025",
  pages="90--102"
}