Skip to content

Releases: andreaconti/depth-on-demand

Models

23 Jan 14:07
Compare
Choose a tag to compare

This release contains a set of wheels to be installed to access the pre-trained models, after having installed the correct version for your platform.

import torch
from depth_on_demand import Model as DoD

device = "cuda:0"
model = DoD(pretrained="scannetv2", device=device)
# images normalized with mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225)
depth = model(
    target=torch.randn(1, 3, 480, 640).to(device), 
    source=torch.randn(1, 3, 480, 640).to(device),
    hints=torch.randn(1, 1, 480, 640).to(device),
    pose_src_tgt=torch.randn(1, 4, 4).to(device),
    intrinsics=torch.randn(1, 2, 3, 3).to(device),
    n_cycles=n_cycles,
)

Data

08 Mar 15:14
Compare
Choose a tag to compare

This release contains test splits for some of the testing datasets to ease the download
and evaluation of the models