Skip to content

federicatopazio/Blood_Cell_Classification

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Artificial Neural Networks and Deep Learning HW1 - AY 2024/2025

Screenshot

Challenge: Blood Smear Classification

This challenge aimed to classify peripheral blood smears into eight categories using deep learning. We explored:

  1. Custom CNNs
  2. Transfer Learning (ImageNet)
  3. Test Time Augmentation (TTA)

Best approach: Transfer Learning with ConvNeXt models.

Data & Augmentation

Dataset: 13,758 images, with 1,799 duplicates removed. Maintaining the original class distribution provided the best results.

Key augmentations:

  • Rotation, flipping
  • Contrast, brightness adjustments
  • Gaussian noise
  • RandAugment pipeline

Performance Boost: VGG models improved from 0.61 → 0.72, ConvNeXt from 0.88 → 0.93.

Models & Training

Top models: ConvNeXt > EfficientNet > VGG.

  • Transfer Learning: Used pre-trained ImageNet weights.
  • TTA: Applied but limited by Codabench constraints.
  • Hyperparameter Tuning: Optimized learning rate, batch size.
  • Best optimizer: NAdam.

Final Results: Accuracy 0.9427, F1 0.9385.

More Info

Refer to the report and notebooks.

Team

About

Blood Classification Neural Network training for university challenge

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors