README.md

metadata

license: apache-2.0
task_categories:
  - image-classification
language:
  - en
  - pt
tags:
  - image-regression
  - image-classification
  - u-tube-manometer
  - gauge-reader
  - water-column
pretty_name: U-Tube Manometer Dataset 2025-2026
size_categories:
  - 10K<n<100K

U-Tube Manometer Dataset (2025–2026)

This dataset contains images of U-tube manometers and is designed for computer vision regression and classification tasks, focusing on reading and interpreting liquid column levels.

• Performance may vary under extreme lighting or occlusion conditions

---

Dataset Summary

The U-Tube Manometer Dataset consists of 20,000 images of U-tube manometers with visible water columns. The dataset is suitable for:

• Image Regression: estimating the numerical value of the water column height
• Image Classification: classifying images based on discrete column values or ranges

Images are organized in a folder structure where the label is encoded in the directory name, representing the measured water column value.

---

Data Structure

The dataset follows a standard split structure:

dataset/
├── train/
│   ├── 10.2/
│   │   ├── image23.png
│   │   └── ...
│   ├── 10.4/
│   └── ...
├── validation/
│   ├── 9.8/
│   └── ...
└── test/
    ├── 10.0/
    └── ...

• Each subfolder name corresponds to the label value (e.g., water column height in centimeters).
• Images inside the folder share the same height value.

---

Image Characteristics

• RGB images
• Varying lighting conditions
• Varying working fluids (water, oil, water–ink mixtures)
• Single U-tube manometer per image
• Visible liquid columns and scale markings

---

License

This dataset is released under the Apache License 2.0.

---

Citation

If you use this dataset in your research, please cite:

@dataset{utm_dataset,
    title     = {U-Tube Manometer Dataset},
    author    = {Santos, Clayton Silva},
    year      = {2026},
    publisher = {Hugging Face},
    url       = {https://huggingface.co/datasets/claytonsds/UTM_DATASET}
}