zaina-ml/white_blood_cell_classification
A PyTorch image classifier that identifies 5 white blood cell types from microscopy images using a fine-tuned DenseNet-121 model.
What's novel
A PyTorch image classifier that identifies 5 white blood cell types from microscopy images using a fine-tuned DenseNet-121 model.
Code Analysis
7 files read · 3 roundsA PyTorch-based image classification pipeline that fine-tunes a pretrained DenseNet-121 backbone to classify White Blood Cells into five categories using weighted sampling and mixed precision training.
Strengths
Excellent modularity with clear separation of concerns (model, dataset, train, evaluate). Implements modern best practices including differential learning rates, gradient accumulation, mixed precision, and robust data loading pipelines. Configuration is centralized and clean.
Weaknesses
Lacks any test suite entirely. Error handling is minimal (no try/except blocks for common failures like missing datasets or corrupted checkpoints). The GPU-transform optimization in dataset.py can be risky if VRAM is constrained.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
6
Contributors
1
Files
12
Active weeks
1
Repository
Language
Jupyter Notebook
Stars
0
Forks
0
License
MIT