File size: 5,762 Bytes
94d9e78
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
 
 
 
 
 
 
 
 
ec31667
 
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
ec31667
94d9e78
 
 
 
 
 
 
 
 
 
ec31667
94d9e78
ec31667
94d9e78
 
ec31667
94d9e78
ec31667
94d9e78
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ec31667
94d9e78
ec31667
94d9e78
 
 
ec31667
 
94d9e78
 
 
 
 
 
 
 
 
 
ec31667
 
94d9e78
ec31667
 
 
 
 
 
 
 
94d9e78
ec31667
94d9e78
ec31667
 
 
94d9e78
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
---
license: mit
pretty_name: Bacteria 2033 Images 33 Types Data
task_categories:
- image-classification
language:
- en
tags:
- bacteria
- microscopy
- microbiology
- biomedical
- medical-imaging
- gram-stain
- computer-vision
- image-classification
- transfer-learning
- digital-twin
size_categories:
- 1K<n<10K
viewer: false
---

# Bacteria 2033 Images 33 Types Data

## Dataset Summary

**Bacteria 2033 Images 33 Types Data** is a microscopy image dataset for bacterial image classification, transfer learning, and biomedical computer vision research. It contains **2,033 RGB microscopy images** representing **33 bacterial types**, with images organized by class label.

The dataset is intended for research on microorganism recognition, clinical microscopy image analysis, deep transfer learning, biological digital twins, and related machine learning workflows.

![Bacteria microscopy examples](https://github.com/user-attachments/assets/7a3044b8-8f1d-427b-8e32-169414c42652)

## Dataset Details

| Field | Description |
|---|---|
| Dataset type | Image classification |
| Modality | RGB microscopy images |
| Number of images | 2,033 |
| Number of classes | 33 bacterial types |
| Annotation | Class labels assigned by laboratory experts |
| Staining | Gram-stained microscopy images |
| Source material | Clinical sample imagery, including blood, urine, and skin samples |
| Repository data format | One folder per bacterial class |
| Download size | Approximately 3.4 GB |
| License | MIT |

## Access and Download

The image archive is distributed through Google Drive:

[Download the dataset archive](https://drive.google.com/file/d/1aR7Dz11wKV3t7awnnnO32UE_37MYF6wX/view?usp=sharing)

After downloading, extract the archive locally. The expected layout is:

```text
Bacteria-2033Images-33Types-dataset/
|-- class_1/
|   |-- image_001.jpg
|   `-- ...
|-- class_2/
|   |-- image_001.jpg
|   `-- ...
`-- ...
```

This folder-per-class structure is compatible with common image-classification loaders such as:

- PyTorch: `torchvision.datasets.ImageFolder`
- TensorFlow/Keras: `tf.keras.utils.image_dataset_from_directory`

## Intended Uses

This dataset may be used for:

- Bacterial species or bacterial type image classification
- Deep transfer learning and feature extraction experiments
- Microorganism digital twin research
- Microscopy image analysis
- Biomedical computer vision benchmarking
- Federated learning and distributed learning experiments where local image folders are used as client data

## Out-of-Scope Uses

This dataset is not intended to be used as a standalone clinical diagnostic system. Models trained on this dataset should not be deployed for clinical decision-making without independent validation, regulatory review, and domain expert oversight.

## Recommended Machine Learning Setup

No official train, validation, or test split is provided in the repository. For reproducible research, users should report:

- The exact train, validation, and test split strategy
- Random seeds
- Image resizing and normalization settings
- Data augmentation methods
- Class balancing or reweighting strategy
- Model architecture and pretrained weights, if any

Because 2,033 images are distributed across 33 classes, users should check for class imbalance and consider stratified splitting, augmentation, weighted losses, or balanced sampling.

## Data Preprocessing

A typical transfer-learning preprocessing workflow is:

1. Extract the archive.
2. Load images with a folder-per-class image loader.
3. Resize images to the model input size.
4. Normalize RGB channels using the selected model's expected preprocessing.
5. Use stratified train, validation, and test splits.

## Limitations

- The dataset size is moderate for 33-class classification, so models may overfit without augmentation or transfer learning.
- Class imbalance may affect evaluation metrics.
- Results may depend strongly on image preprocessing, magnification, staining variation, and split strategy.
- The dataset should be validated on external microscopy data before use in any applied biomedical setting.

## Citation Requirement

If you use this dataset, derivative labels, trained models, figures, benchmarks, or results produced from this dataset, you **must cite both** of the following papers:

1. M. B. Jamshidi, D. T. Hoang, D. N. Nguyen, D. Niyato, and M. E. Warkiani, "Revolutionizing biological digital twins: Integrating internet of bio-nano things, convolutional neural networks, and federated learning," *Computers in Biology and Medicine*, vol. 189, p. 109970, 2025.

2. M. B. Jamshidi, S. Sargolzaei, S. Foorginezhad, and O. Moztarzadeh, "Metaverse and microorganism digital twins: A deep transfer learning approach," *Applied Soft Computing*, vol. 147, p. 110798, 2023.

### BibTeX

```bibtex
@article{jamshidi2025revolutionizing,
  title={Revolutionizing biological digital twins: Integrating internet of bio-nano things, convolutional neural networks, and federated learning},
  author={Jamshidi, M. B. and Hoang, D. T. and Nguyen, D. N. and Niyato, D. and Warkiani, M. E.},
  journal={Computers in Biology and Medicine},
  volume={189},
  pages={109970},
  year={2025},
  publisher={Elsevier}
}

@article{jamshidi2023metaverse,
  title={Metaverse and microorganism digital twins: A deep transfer learning approach},
  author={Jamshidi, M. B. and Sargolzaei, S. and Foorginezhad, S. and Moztarzadeh, O.},
  journal={Applied Soft Computing},
  volume={147},
  pages={110798},
  year={2023},
  publisher={Elsevier}
}
```

## License

This dataset is released under the MIT License.

## Maintainer

Mohammad Behdad Jamshidi

- Hugging Face: [MBJamshidi](https://huggingface.co/MBJamshidi)
- GitHub: [MBJamshidi](https://github.com/MBJamshidi)