raoyongming/DynamicViT
[NeurIPS 2021] [T-PAMI] DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification
This project helps machine learning engineers and researchers accelerate their computer vision models. By dynamically removing less important parts of an image as it's processed, it takes existing image or video data and outputs faster, more efficient models for tasks like image classification, object detection, and semantic segmentation. This is ideal for those working with large datasets or deploying models to resource-constrained environments.
651 stars. No commits in the last 6 months.
Use this if you need to significantly speed up your image and video processing models (like Vision Transformers or CNNs) without losing much accuracy.
Not ideal if your primary concern is absolute peak accuracy and you have no constraints on computational resources or inference speed.
Stars
651
Forks
80
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Jul 11, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/raoyongming/DynamicViT"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Jittor/jittor
Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
zhanghang1989/ResNeSt
ResNeSt: Split-Attention Networks
berniwal/swin-transformer-pytorch
Implementation of the Swin Transformer in PyTorch.
NVlabs/FasterViT
[ICLR 2024] Official PyTorch implementation of FasterViT: Fast Vision Transformers with...
ViTAE-Transformer/ViTPose
The official repo for [NeurIPS'22] "ViTPose: Simple Vision Transformer Baselines for Human Pose...