Awesome

Cellshape-cloud is an easy-to-use tool to analyse the shapes of cells using deep learning and, in particular, graph-neural networks. The tool provides the ability to train popular graph-based autoencoders on point cloud data of 2D and 3D single cell masks as well as providing pre-trained networks for inference.

To install

pip install cellshape-cloud

Usage

Basic Usage

import torch
from cellshape_cloud import CloudAutoEncoder

model = CloudAutoEncoder(num_features=128, 
                         k=20,
                         encoder_type="dgcnn",
                         decoder_type="foldingnet")

points = torch.randn(1, 2048, 3)

recon, features = model(points)

To train an autoencoder on a set of point clouds created using cellshape-helper:

import torch
from torch.utils.data import DataLoader

import cellshape_cloud as cloud
from cellshape_cloud.vendor.chamfer_distance import ChamferLoss


input_dir = "path/to/pointcloud/files/"
batch_size = 16
learning_rate = 0.0001
num_epochs = 1
output_dir = "path/to/save/output/"

model = cloud.CloudAutoEncoder(num_features=128, 
                         k=20,
                         encoder_type="dgcnn",
                         decoder_type="foldingnet")

dataset = cloud.PointCloudDataset(input_dir)

dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=True)

criterion = ChamferLoss()

optimizer = torch.optim.Adam(
    model.parameters(),
    lr=learning_rate * 16 / batch_size,
    betas=(0.9, 0.999),
    weight_decay=1e-6,
)

cloud.train(model, dataloader, num_epochs, criterion, optimizer, output_dir)

Parameters

num_features: int.
The size of the latent space of the autoencoder.
k: int.
The number of neightbours to use in the k-nearest-neighbours graph construction.
encoder_type: str.
The type of encoder: 'foldingnet' or 'dgcnn'
decoder_type: str.
The type of decoder: 'foldingnet' or 'dgcnn'

References

[1] An Tao, 'Unsupervised Point Cloud Reconstruction for Classific Feature Learning', GitHub Repo, 2020