Reinventing 2D Convolutions for 3D Images (arXiv)
- ACS convolution aims at a plug-and-play replacement of standard 3D convolution, for 3D medical images.
- ACS convolution enables 2D-to-3D transfer learning, which consistently provides significant performance boost in our experiments.
- Even without pretraining, ACS convolution is comparable to or even better than 3D convolution, with smaller model size and less computation.
acsconv
the core implementation of ACS convolution, including the operators, models, and 2D-to-3D/ACS model converters.operators
: include ACSConv, SoftACSConv and Conv2_5d.converters
: include converters which convert 2D models to 3d/ACS/Conv2_5d counterparts.models
: Native ACS models.
experiments
the scripts to run experiments.mylib
: the lib for running the experiments.poc
: the scripts to run proof-of-concept experiments.lidc
: the scripts to run LIDC-IDRI experiments.
from torchvision.models import resnet18
from acsconv.converters import ACSConverter
# model_2d is a standard pytorch 2D model
model_2d = resnet18(pretrained=True)
B, C_in, H, W = (1, 3, 64, 64)
input_2d = torch.rand(B, C_in, H, W)
output_2d = model_2d(input_2d)
model_3d = ACSConverter(model_2d)
# once converted, model_3d is using ACSConv and capable of processing 3D volumes.
B, C_in, D, H, W = (1, 3, 64, 64, 64)
input_3d = torch.rand(B, C_in, D, H, W)
output_3d = model_3d(input_3d)
from acsconv.operators import ACSConv, SoftACSConv
x = torch.rand(batch_size, 3, D, H, W)
# ACSConv to process 3D volumnes
conv = ACSConv(in_channels=3, out_channels=10, kernel_size=3, padding=1)
out = conv(x)
# SoftACSConv to process 3D volumnes
conv = SoftACSConv(in_channels=3, out_channels=10, kernel_size=3, padding=1)
out = conv(x)
from acsconv.models.acsunet import ACSUnet
unet_3d = ACSUnet(num_classes=3)
B, C_in, D, H, W = (1, 3, 64, 64, 64)
input_3d = torch.rand(B, C_in, D, H, W)
output_3d = unet_3d(input_3d)
[WIP] More code is coming soon, stay tuned!
- More document
- Memory-efficient implementation
- More pretrained models (ours / other open source projects)