Datasets:
image
imagewidth (px) 150
5.5k
|
---|
Dataset Card for Bacterial Morphology Classification Dataset
Dataset Summary
This dataset contains images of bacteria categorized into three morphological classes: cocci, bacilli, and spirilla. It is designed for training, validating, and testing machine learning models for bacterial morphology classification.
Dataset Details
Dataset Description
This dataset was created to support research in microbiology and machine learning, focusing on bacterial morphology classification. Images in the dataset are labeled into three classes: cocci, bacilli, and spirilla.
- Dataset Title: Bacterial Morphology Classification
- Dataset Curator: Yola Charara, University of Michigan-Flint, M.S Data Science-Student
- Dataset Version: The data was collected from September 28, 2024 till October 1, 2024.
- Data Card Author: Yola Charara, University of Michigan-Flint, M.S Data Science-Student
- Data Card Version: December 6, 2024
- License: CC BY 4.0
Uses
Direct Use
The dataset is intended for training and evaluating machine learning models to classify bacteria by morphology, supporting tasks in microbiology and biomedical research.
Out-of-Scope Use
The dataset is not suitable for applications outside bacterial morphology classification. Misuse could include attempts to infer specific bacterial species or other unrelated properties.
Dataset Structure
The dataset is divided into training, validation, and test sets. Images in the training and validation sets are grouped by class, while the test set contains images without class-based folders.
Dataset Creation
Curation Rationale
The dataset was created to facilitate automated bacterial classification, reducing reliance on manual microscopy analysis and accelerating research workflows.
Source Data
Data Collection and Processing
Images were curated from the python library bing-downloader which downloads a bulk of images from bing images. Preprocessing included resizing, normalizing, and re-saving all .jpg
images in RGB mode to ensure compatibility and resolve encoding issues.
Who are the source data producers?
The images were sourced from microbiological research or publicly available datasets. Specific data producers are not disclosed.
Annotations
Annotation process
Each image was labeled based on its morphological class by graduate students.
Personal and Sensitive Information
The dataset does not contain any personal or sensitive information.
Bias, Risks, and Limitations
The dataset is limited to three morphological classes and may not generalize well to other bacterial shapes or conditions. There may be inherent biases in the dataset depending on the sources of the images.
Recommendations
Users should test models on other datasets to ensure generalizability and reduce bias, as this dataset may not cover all variations. Additional data from diverse sources is recommended to improve model robustness.
Glossary
Cocci: Round-shaped bacteria.
Bacilli: Rod-shaped bacteria.
Spirilla: Spiral-shaped bacteria.
More Information
Codalab link for this dataset: https://codalab.lisn.upsaclay.fr/competitions/20680?secret_key=b04da988-294a-4491-831c-3f4b55529a0b
Dataset Card Contact
email: yolac@umich.edu
- Downloads last month
- 63