wildphoton/RandConv

Code for ICLR2021 paper "Robust and Generalizable Visual Representation Learning via Random Convolutions"

26
/ 100
Experimental

This project helps machine learning practitioners improve the reliability and transferability of their image recognition models, especially when dealing with new or noisy visual data. By training models using a technique called Random Convolutions, it takes in image datasets (like digits or PACS images) and produces more robust visual representations. Data scientists and ML engineers working on computer vision tasks will find this useful.

No commits in the last 6 months.

Use this if you need your image recognition models to perform consistently well even when faced with variations or corruptions in image data that weren't present during initial training.

Not ideal if you are looking for a general-purpose image labeling or object detection tool rather than a method for improving the foundational robustness of your visual models.

computer-vision image-recognition model-robustness domain-generalization deep-learning
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 8 / 25
Community 10 / 25

How are scores calculated?

Stars

69

Forks

6

Language

Python

License

Last pushed

May 10, 2021

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/wildphoton/RandConv"

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