Batching
-In deep learning, every optimization step operates on multiple input examples for robust training. Thus, efficient batching is crucial. For image inputs, batching is straighforward; N images are resized to the same height and width and stacked as a 4 dimensional tensor of shape N x 3 x H x W. For meshes, batching is less straighforward.
In deep learning, every optimization step operates on multiple input examples for robust training. Thus, efficient batching is crucial. For image inputs, batching is straightforward; N images are resized to the same height and width and stacked as a 4 dimensional tensor of shape N x 3 x H x W. For meshes, batching is less straightforward.
Batch modes for meshes
Assume you want to construct a batch containing two meshes, with mesh1 = (v1: V1 x 3, f1: F1 x 3) containing V1 vertices and F1 faces, and mesh2 = (v2: V2 x 3, f2: F2 x 3) with V2 (!= V1) vertices and F2 (!= F1) faces. The Meshes data structure provides three different ways to batch heterogeneous meshes. If meshes = Meshes(verts = [v1, v2], faces = [f1, f2]) is an instantiation of the data structure, then
Use cases for batch modes
-The need for different mesh batch modes is inherent to the way pytorch operators are implemented. To fully utilize the optimized pytorch ops, the Meshes data structure allows for efficient conversion between the different batch modes. This is crucial when aiming for a fast and efficient training cycle. An example of this is Mesh R-CNN. Here, in the same forward pass different parts of the network assume different inputs, which are computed by converting between the different batch modes. In particular, vert_align assumes a padded input tensor while immediately after graph_conv assumes a packed input tensor.
+The need for different mesh batch modes is inherent to the way PyTorch operators are implemented. To fully utilize the optimized PyTorch ops, the Meshes data structure allows for efficient conversion between the different batch modes. This is crucial when aiming for a fast and efficient training cycle. An example of this is Mesh R-CNN. Here, in the same forward pass different parts of the network assume different inputs, which are computed by converting between the different batch modes. In particular, vert_align assumes a padded input tensor while immediately after graph_conv assumes a packed input tensor.
