shiranzada/pure-noise
Official implementation for "Pure Noise to the Rescue of Insufficient Data: Improving Imbalanced Classification by Training on Random Noise Images" https://arxiv.org/abs/2112.08810
This project helps image recognition systems perform better when you have very few examples for some categories. It takes your existing image datasets, especially those with an uneven number of pictures per category, and incorporates synthetic noise images during training. The output is a more robust image classification model that can accurately identify items even in underrepresented categories. This is for anyone who builds or uses image classification models, such as researchers, data scientists, or machine learning engineers, who struggle with imbalanced image datasets.
No commits in the last 6 months.
Use this if you are working with image classification tasks and your model struggles because some categories have significantly fewer training examples than others.
Not ideal if your problem is not image classification or if your datasets are already perfectly balanced and extensive across all categories.
Stars
15
Forks
—
Language
—
License
—
Category
Last pushed
Jun 11, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/shiranzada/pure-noise"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
CAREamics/careamics
A deep-learning library for denoising images using Noise2Void and friends (CARE, PN2V, HDN...
yu4u/noise2noise
An unofficial and partial Keras implementation of "Noise2Noise: Learning Image Restoration...
rgeirhos/texture-vs-shape
Pre-trained models, data, code & materials from the paper "ImageNet-trained CNNs are biased...
NICALab/SUPPORT
Accurate denoising of voltage imaging data through statistically unbiased prediction, Nature Methods.
jaewon-lee-b/lte
Local Texture Estimator for Implicit Representation Function, in CVPR 2022