voxel51/reconstruction-error-ratios
Estimate dataset difficulty and detect label mistakes using reconstruction error ratios!
This tool helps machine learning engineers and data scientists quickly assess the quality and difficulty of their image classification datasets. By analyzing images and their associated labels, it provides scores that reveal how challenging a dataset is and points out potential labeling errors. You input your image classification dataset, and it outputs overall dataset difficulty, class-level difficulty, and a list of potentially mislabeled images.
No commits in the last 6 months.
Use this if you need to understand the inherent difficulty of your image classification task or want to efficiently find and fix mistakes in your dataset labels.
Not ideal if your dataset does not consist of images with classification labels or if you are not working with computer vision models.
Stars
28
Forks
—
Language
Python
License
—
Category
Last pushed
Jan 10, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/voxel51/reconstruction-error-ratios"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Cloud-CV/EvalAI
:cloud: :rocket: :bar_chart: :chart_with_upwards_trend: Evaluating state of the art in AI
fireindark707/Python-Schema-Matching
A python tool using XGboost and sentence-transformers to perform schema matching task on tables.
graphbookai/graphbook
Visual AI development framework for training and inference of ML models, scaling pipelines, and...
visual-layer/fastdup
fastdup is a powerful, free tool designed to rapidly generate valuable insights from image and...
github/CodeSearchNet
Datasets, tools, and benchmarks for representation learning of code.