shikishima-TasakiLab/Involution-PyTorch

Unofficial PyTorch reimplemention of the paper "Involution: Inverting the Inherence of Convolution for Visual Recognition" [CVPR 2021].

33
/ 100
Emerging

This project offers an alternative to standard convolution layers used in deep learning models for visual recognition tasks. It helps researchers and practitioners experimenting with novel neural network architectures by providing an Involution2d module. This module takes in multi-channel image data (tensors) and produces processed feature maps, which can then be used in subsequent layers of a computer vision model. This is for AI/ML researchers and deep learning engineers building and evaluating custom computer vision models.

No commits in the last 6 months.

Use this if you are a deep learning researcher or engineer exploring new architectural components for computer vision models, specifically looking to replace or augment traditional convolutional layers.

Not ideal if you are looking for a pre-trained model or a high-level API for common computer vision tasks without delving into custom network architectures.

deep-learning-research neural-network-architecture computer-vision image-recognition pytorch-development
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 11 / 25

How are scores calculated?

Stars

21

Forks

3

Language

C++

License

Last pushed

Jul 13, 2021

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/shikishima-TasakiLab/Involution-PyTorch"

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