Vision Mamba - Pytorch
Project description
Vision Mamba
Implementation of Vision Mamba from the paper: "Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model" It's 2.8x faster than DeiT and saves 86.8% GPU memory when performing batch inference to extract features on high-res images.
Installation
pip install vision-mamba
Usage
import torch
from vision_mamba import Vim
# Forward pass
x = torch.randn(1, 3, 224, 224) # Input tensor with shape (batch_size, channels, height, width)
# Model
model = Vim(
dim=256, # Dimension of the transformer model
heads=8, # Number of attention heads
dt_rank=32, # Rank of the dynamic routing matrix
dim_inner=256, # Inner dimension of the transformer model
d_state=256, # Dimension of the state vector
num_classes=1000, # Number of output classes
image_size=224, # Size of the input image
patch_size=16, # Size of each image patch
channels=3, # Number of input channels
dropout=0.1, # Dropout rate
depth=12, # Depth of the transformer model
)
# Forward pass
out = model(x) # Output tensor from the model
print(out.shape) # Print the shape of the output tensor
print(out) # Print the output tensor
Citation
@misc{zhu2024vision,
title={Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model},
author={Lianghui Zhu and Bencheng Liao and Qian Zhang and Xinlong Wang and Wenyu Liu and Xinggang Wang},
year={2024},
eprint={2401.09417},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
License
MIT
Todo
- Create training script for imagenet
- Create a visual mamba for facial recognition
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
vision_mamba-0.0.8.tar.gz
(5.4 kB
view hashes)
Built Distribution
Close
Hashes for vision_mamba-0.0.8-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 63facbbdd1941964f6ea687bd299a14bace88dd9355240c3e65be0f7f4164258 |
|
MD5 | a5dd801005445a3df3c65c558e13642d |
|
BLAKE2b-256 | 47073314bd02c21f71fed676012d40f8c75c03136028e534c340e5fcfa18465b |