mirrors/mmsegmentation
1
0
Fork 0
mirror of https://github.com/open-mmlab/mmsegmentation.git synced 2025-06-03 22:03:48 +08:00
Code Issues Projects Releases Wiki Activity

[Enhancement]Replace numpy ascontiguousarray with torch contiguous to speed-up (#2604)

Browse Source 
## Motivation

Original motivation was after [MMDetection PR
#9533](https://github.com/open-mmlab/mmdetection/pull/9533)

With several experiments I found out that if a ndarray is contiguous,
numpy.transpose + torch.contiguous perform better, while if not, then
use numpy.ascontiguousarray + numpy.transpose

## Modification

Replace numpy.ascontiguousarray with torch.contiguous in
[PackSegInputs](https://github.com/open-mmlab/mmsegmentation/blob/1.x/mmseg/datasets/transforms/formatting.py)

Co-authored-by: MeowZheng <meowzheng@outlook.com>
This commit is contained in:
CSH 2023-02-15 19:02:00 +08:00 committed by GitHub
parent 9b8e8b730c
commit 2e27f8b678
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
mmseg/datasets/transforms/formatting.py
View File

@ -63,8 +63,12 @@ class PackSegInputs(BaseTransform):
img = results['img']
if len(img.shape) < 3:
img = np.expand_dims(img, -1)
img = np.ascontiguousarray(img.transpose(2, 0, 1))
packed_results['inputs'] = to_tensor(img)
if not img.flags.c_contiguous:
img = to_tensor(np.ascontiguousarray(img.transpose(2, 0, 1)))
else:
img = img.transpose(2, 0, 1)
img = to_tensor(img).contiguous()
packed_results['inputs'] = img
data_sample = SegDataSample()
if 'gt_seg_map' in results: