migonsa/analog_clocks_combinations_for_finetuning

Analog Clocks Combinations Dataset for Finetuning

This repository hosts a collection of 43,200 high-quality, synthetic images of analog clocks, generated for every possible hour, minute, and second in a 12-hour cycle, and for each of three clock types:

1. Base: normal clocks.
2. Distorted: dial with distorted shape.
3. Modified hands: hands with the same thickness and with an arrow.

The data is useful for training and benchmarking computer vision models on tasks like time recognition, synthetic recognition, and image understanding. The images are used in the experiments in this preprint: https://arxiv.org/abs/2505.10862

Dataset Structure

Each entry in the dataset contains the following fields:

• type: (string) The type of clock (base, distorted, or modified_hands).
• question: (string) Always "What time is shown on the clock in the given image?"
• answer: (string) Time in the image in HH:MM:SS format.
• image: (image file) Image of the analog clock.

Usage

You can load this dataset using the Hugging Face Datasets library:

from datasets import load_dataset
dataset = load_dataset('migonsa/analog_clocks_combinations_for_finetuning', split='train')

Source and Citation

If you use this dataset, please cite the preprint:

Experiments using these images: https://arxiv.org/abs/2505.10862

License

CC BY 4.0 — You are free to use, share, and adapt with attribution.

---

Considerations & Limitations

• The dataset is synthetic and may not represent real-world clock image variability (lighting, background, occlusion, etc.).
• All images generated in a controlled environment.

---