AlexNet Fine-Tuned on SIPaKMeD
This repository provides a fine-tuned AlexNet model trained on the SIPaKMeD cervical cytology image dataset. The model is designed to classify cervical cell images into multiple morphological categories based on visual characteristics observed in Pap smear samples.
Model Overview
- Architecture: AlexNet
- Framework: PyTorch
- Training Type: Fine-tuning from pretrained weights
- Input Resolution: 224 × 224 RGB images
- Output: Multi-class cervical cell classification
The model learns hierarchical feature representations capturing cellular shape, texture, and structural variations relevant to cytological analysis.
Dataset
The SIPaKMeD dataset consists of cervical cell images categorized into distinct classes representing different cell morphologies commonly observed in Pap smear analysis. The dataset includes both normal and abnormal cell types and is widely used for benchmarking cervical cytology classification methods.
Classes
The model predicts the following cervical cell categories:
- Superficial–Intermediate
- Parabasal
- Koilocytotic
- Dyskeratotic
- Metaplastic
Intended Use
This model is intended for research and experimental purposes, including:
- Benchmarking cervical cytology classification methods
- Feature analysis and visualization studies
- Model comparison and ablation experiments
- Integration with explainable AI techniques
Model Outputs
The model outputs class probabilities for each cervical cell category. The predicted label corresponds to the class with the highest probability score.
Notes
- The model was trained using standardized image preprocessing including resizing and normalization.
- Performance may vary when applied to datasets with different staining protocols or imaging conditions.
- Users are encouraged to fine-tune or recalibrate the model when transferring to new datasets.
- Downloads last month
- 3