MobileNetV3-Large for Bacterial Colony Classification

This model is a fine-tuned version of MobileNetV3-Large on the DIBaS (Digital Image of Bacterial Species) dataset for classifying bacterial colony images into 33 species.

Model Description

Model Architecture: MobileNetV3-Large (pretrained on ImageNet)
Task: Multi-class image classification (33 bacterial species)
Dataset: DIBaS - 660+ microscopy images of bacterial colonies
Framework: PyTorch + timm

Performance

Metric	Value
Validation Accuracy	95.45%
Macro F1-Score	0.954
Parameters	4.24M
Model Size	16.4 MB
GPU Latency	4.39 ms (RTX 4070 SUPER)
CPU Latency	20.33 ms

Comparison with Other Architectures

Model	Params (M)	Val Accuracy
MobileNetV3-Large	4.24	95.45%
ResNet50	23.58	93.94%
EfficientNet-B3	10.75	92.42%
ViT-Tiny	5.53	78.79%

Intended Use

This model is designed for:

Automated bacterial species identification from colony images
Research in clinical microbiology
Educational purposes in microbiology labs
Mobile/edge deployment for rapid identification

Training Details

Optimizer: AdamW (lr=1e-3, weight_decay=1e-4)
Epochs: 20
Batch Size: 32
Image Size: 224×224
Augmentation: RandomResizedCrop, HorizontalFlip
Hardware: NVIDIA RTX 4070 SUPER
Mixed Precision: Enabled (AMP)
Train/Val/Test Split: 70/20/10 (stratified, seed=42)

How to Use

With timm (Recommended)

import timm
import torch
from PIL import Image
from torchvision import transforms

# Load model
model = timm.create_model('mobilenetv3_large_100', pretrained=False, num_classes=33)
state_dict = torch.load('pytorch_model.pth', map_location='cpu')
model.load_state_dict(state_dict['model_state'])
model.eval()

# Preprocessing
transform = transforms.Compose([
    transforms.Resize(256),
    transforms.CenterCrop(224),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])

# Inference
image = Image.open('bacteria_image.jpg').convert('RGB')
input_tensor = transform(image).unsqueeze(0)

with torch.no_grad():
    outputs = model(input_tensor)
    predicted_class = outputs.argmax(dim=1).item()
    confidence = torch.softmax(outputs, dim=1)[0, predicted_class].item()

print(f"Predicted class: {CLASS_NAMES[predicted_class]}")
print(f"Confidence: {confidence:.2%}")

Class Labels

CLASS_NAMES = [
    "Acinetobacter_baumannii", "Actinomyces_israelii", "Bacteroides_fragilis",
    "Bifidobacterium_spp", "Candida_albicans", "Clostridium_perfringens",
    "Enterococcus_faecalis", "Enterococcus_faecium", "Escherichia_coli",
    "Fusobacterium", "Lactobacillus_casei", "Lactobacillus_crispatus",
    "Lactobacillus_delbrueckii", "Lactobacillus_gasseri", "Lactobacillus_jensenii",
    "Lactobacillus_johnsonii", "Lactobacillus_paracasei", "Lactobacillus_plantarum",
    "Lactobacillus_reuteri", "Lactobacillus_rhamnosus", "Lactobacillus_salivarius",
    "Listeria_monocytogenes", "Micrococcus_spp", "Neisseria_gonorrhoeae",
    "Porphyromonas_gingivalis", "Propionibacterium_acnes", "Proteus",
    "Pseudomonas_aeruginosa", "Staphylococcus_aureus", "Staphylococcus_epidermidis",
    "Staphylococcus_saprophyticus", "Streptococcus_agalactiae", "Veillonella"
]

Limitations

Trained on single laboratory/microscope setup (DIBaS dataset)
May not generalize to different imaging conditions or staining methods
Some morphologically similar species (e.g., Staphylococcus variants) may be confused
Not validated for clinical diagnostic use

Citation

@inproceedings{hoflaz2025bacterial,
  title={Lightweight CNNs Outperform Vision Transformers for Bacterial Colony Classification},
  author={Hoflaz, Ibrahim},
  booktitle={IEEE Conference},
  year={2025}
}

References

DIBaS Dataset: http://misztal.edu.pl/software/databases/dibas/
Zieliński, B., et al. "Deep learning approach to bacterial colony classification." PloS one 12.9 (2017): e0184554.

Model Card Contact

GitHub: ihoflaz/bacterial-colony-classification
Author: Ibrahim Hoflaz

Downloads last month: 14