Inference time too long on ade20k validation set (HRNet + OCR)

HRNet / HRNet-Semantic-Segmentation

The OCR approach is rephrased as Segmentation Transformer: https://arxiv.org/abs/1909.11065. This is an official implementation of semantic segmentation for HRNet. https://arxiv.org/abs/1908.07919

Other

3.09k stars 683 forks source link

Inference time too long on ade20k validation set (HRNet + OCR) #204

Closed lubovbyc closed 3 years ago

lubovbyc commented 3 years ago

dataset: ade20k validation set device: single gpu (V100) time: ~5h

Multi-scale inference seems too slow on this dataset. Is it normal?

PkuRainBow commented 3 years ago

@lubovbyc I guess this is normal as this implementation does not support multi-GPU testing. Please try the openseg.pytorch for multiscale evaluation with multi-GPUs.