From 0fa7c95fd93c38da2a5e4c0567b90538bbb35d3d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E8=B0=A2=E6=98=95=E8=BE=B0?= Date: Thu, 7 Jul 2022 19:24:02 +0800 Subject: [PATCH] update erfnet results (#1744) --- configs/erfnet/README.md | 8 +++++--- configs/erfnet/erfnet.yml | 6 +++--- 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/configs/erfnet/README.md b/configs/erfnet/README.md index bcb61d3d6..b7b83df54 100644 --- a/configs/erfnet/README.md +++ b/configs/erfnet/README.md @@ -41,12 +41,14 @@ Semantic segmentation is a challenging task that addresses most of the perceptio ### Cityscapes -| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download | -| ------ | -------- | --------- | ------: | -------- | -------------- | ----: | ------------- | ---------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | -| ERFNet | ERFNet | 512x1024 | 160000 | 6.04 | 15.26 | 71.08 | 72.6 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20211126_082056-03d333ed.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20211126_082056.log.json) | +| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download | +| ------ | -------- | --------- | ------: | -------- | -------------- | ---: | ------------- | ---------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | +| ERFNet | ERFNet | 512x1024 | 160000 | 6.04 | 15.26 | 72.5 | 74.75 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20220704_162145-dc90157a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20220704_162145.log.json) | Note: - The model is trained from scratch. - Last deconvolution layer in the [original paper](https://github.com/Eromera/erfnet_pytorch/blob/master/train/erfnet.py#L123) is replaced by a naive `FCNHead` decoder head and a bilinear upsampling layer, found more effective and efficient. + +- This model performance is sensitive to the seed values used, please refer to the log file for the specific settings of the seed. If you choose a different seed, the results might differ from the table results. diff --git a/configs/erfnet/erfnet.yml b/configs/erfnet/erfnet.yml index e4c34f9c5..820f5ed6c 100644 --- a/configs/erfnet/erfnet.yml +++ b/configs/erfnet/erfnet.yml @@ -31,7 +31,7 @@ Models: - Task: Semantic Segmentation Dataset: Cityscapes Metrics: - mIoU: 71.08 - mIoU(ms+flip): 72.6 + mIoU: 72.5 + mIoU(ms+flip): 74.75 Config: configs/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes.py - Weights: https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20211126_082056-03d333ed.pth + Weights: https://download.openmmlab.com/mmsegmentation/v0.5/erfnet/erfnet_fcn_4x4_512x1024_160k_cityscapes/erfnet_fcn_4x4_512x1024_160k_cityscapes_20220704_162145-dc90157a.pth