richardaecn/class-balanced-loss

Class-Balanced Loss Based on Effective Number of Samples. CVPR 2019

45
/ 100
Emerging

When training image classification models, this project helps scientists and machine learning practitioners overcome the common problem of imbalanced datasets, where some categories have many more examples than others. It takes a dataset like CIFAR or iNaturalist and applies a special 'class-balanced' weighting to the training process. The outcome is a more accurate classification model, especially for those rare categories.

616 stars. No commits in the last 6 months.

Use this if your image classification model struggles to accurately identify objects in categories that have very few examples compared to others.

Not ideal if your dataset is already perfectly balanced across all object categories or if you are not working with image classification tasks.

image-classification imbalanced-data computer-vision machine-learning-training
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 19 / 25

How are scores calculated?

Stars

616

Forks

69

Language

Python

License

MIT

Last pushed

Aug 29, 2021

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/richardaecn/class-balanced-loss"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.