[Feature] Provide URLs of STDC, Segmenter and Twins pretrained models (open-mmlab#1357)

MengzhangLI · web-flow · commit 8f33d68b42ef · 2022-03-09T13:19:41.000+08:00
diff --git a/configs/_base_/models/segmenter_vit-b16_mask.py b/configs/_base_/models/segmenter_vit-b16_mask.py
@@ -1,8 +1,9 @@
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_base_p16_384_20220308-96dfe169.pth'  # noqa
 # model settings
 backbone_norm_cfg = dict(type='LN', eps=1e-6, requires_grad=True)
 model = dict(
     type='EncoderDecoder',
-    pretrained='pretrain/vit_base_p16_384.pth',
+    pretrained=checkpoint,
     backbone=dict(
         type='VisionTransformer',
         img_size=(512, 512),
diff --git a/configs/_base_/models/twins_pcpvt-s_fpn.py b/configs/_base_/models/twins_pcpvt-s_fpn.py
@@ -1,12 +1,13 @@
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_small_20220308-e638c41c.pth'  # noqa
+
 # model settings
 backbone_norm_cfg = dict(type='LN')
 norm_cfg = dict(type='SyncBN', requires_grad=True)
 model = dict(
     type='EncoderDecoder',
     backbone=dict(
         type='PCPVT',
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_small.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         in_channels=3,
         embed_dims=[64, 128, 320, 512],
         num_heads=[1, 2, 5, 8],
diff --git a/configs/_base_/models/twins_pcpvt-s_upernet.py b/configs/_base_/models/twins_pcpvt-s_upernet.py
@@ -1,12 +1,13 @@
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_small_20220308-e638c41c.pth'  # noqa
+
 # model settings
 backbone_norm_cfg = dict(type='LN')
 norm_cfg = dict(type='SyncBN', requires_grad=True)
 model = dict(
     type='EncoderDecoder',
     backbone=dict(
         type='PCPVT',
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_small.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         in_channels=3,
         embed_dims=[64, 128, 320, 512],
         num_heads=[1, 2, 5, 8],
diff --git a/configs/segmenter/README.md b/configs/segmenter/README.md
@@ -33,9 +33,9 @@ Image segmentation is often ambiguous at the level of individual image patches a
 
 ## Usage
 
-To use the pre-trained ViT model from [Segmenter](https://github.com/rstrudel/segmenter), it is necessary to convert keys.
+We have provided pretrained models converted from [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106).
 
-We provide a script [`vitjax2mmseg.py`](../../tools/model_converters/vitjax2mmseg.py) in the tools directory to convert the key of models from [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106) to MMSegmentation style.
+If you want to convert keys on your own to use the pre-trained ViT model from [Segmenter](https://github.com/rstrudel/segmenter), we also provide a script [`vitjax2mmseg.py`](../../tools/model_converters/vitjax2mmseg.py) in the tools directory to convert the key of models from [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106) to MMSegmentation style.
 
 ```shell
 python tools/model_converters/vitjax2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
diff --git a/configs/segmenter/segmenter_vit-l_mask_8x1_512x512_160k_ade20k.py b/configs/segmenter/segmenter_vit-l_mask_8x1_512x512_160k_ade20k.py
@@ -3,9 +3,10 @@
     '../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
     '../_base_/schedules/schedule_160k.py'
 ]
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_large_p16_384_20220308-d4efb41d.pth'  # noqa
 
 model = dict(
-    pretrained='pretrain/vit_large_p16_384.pth',
+    pretrained=checkpoint,
     backbone=dict(
         type='VisionTransformer',
         img_size=(640, 640),
diff --git a/configs/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py b/configs/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py
@@ -4,9 +4,11 @@
     '../_base_/schedules/schedule_160k.py'
 ]
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_small_p16_384_20220308-410f6037.pth'  # noqa
+
 backbone_norm_cfg = dict(type='LN', eps=1e-6, requires_grad=True)
 model = dict(
-    pretrained='pretrain/vit_small_p16_384.pth',
+    pretrained=checkpoint,
     backbone=dict(
         img_size=(512, 512),
         embed_dims=384,
diff --git a/configs/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k.py b/configs/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k.py
@@ -4,8 +4,10 @@
     '../_base_/schedules/schedule_160k.py'
 ]
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_tiny_p16_384_20220308-cce8c795.pth'  # noqa
+
 model = dict(
-    pretrained='pretrain/vit_tiny_p16_384.pth',
+    pretrained=checkpoint,
     backbone=dict(embed_dims=192, num_heads=3),
     decode_head=dict(
         type='SegmenterMaskTransformerHead',
diff --git a/configs/stdc/README.md b/configs/stdc/README.md
@@ -35,9 +35,9 @@ BiSeNet has been proved to be a popular two-stream network for real-time segment
 
 ## Usage
 
-To use original repositories' [ImageNet Pretrained STDCNet Weights](https://drive.google.com/drive/folders/1wROFwRt8qWHD4jSo8Zu1gp1d6oYJ3ns1) , it is necessary to convert keys.
+We have provided [ImageNet Pretrained STDCNet Weights](https://drive.google.com/drive/folders/1wROFwRt8qWHD4jSo8Zu1gp1d6oYJ3ns1) models converted from [official repo](https://github.com/MichaelFan01/STDC-Seg).
 
-We provide a script [`stdc2mmseg.py`](../../tools/model_converters/stdc2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/MichaelFan01/STDC-Seg) to MMSegmentation style.
+If you want to convert keys on your own to use official repositories' pre-trained models, we also provide a script [`stdc2mmseg.py`](../../tools/model_converters/stdc2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/MichaelFan01/STDC-Seg) to MMSegmentation style.
 
 ```shell
 python tools/model_converters/stdc2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH} ${STDC_TYPE}
diff --git a/configs/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes.py b/configs/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes.py
@@ -1,6 +1,6 @@
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/stdc/stdc1_20220308-5368626c.pth'  # noqa
 _base_ = './stdc1_512x1024_80k_cityscapes.py'
 model = dict(
     backbone=dict(
         backbone_cfg=dict(
-            init_cfg=dict(
-                type='Pretrained', checkpoint='./pretrained/stdc1.pth'))))
+            init_cfg=dict(type='Pretrained', checkpoint=checkpoint))))
diff --git a/configs/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes.py b/configs/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes.py
@@ -1,6 +1,6 @@
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/stdc/stdc2_20220308-7dbd9127.pth'  # noqa
 _base_ = './stdc2_512x1024_80k_cityscapes.py'
 model = dict(
     backbone=dict(
         backbone_cfg=dict(
-            init_cfg=dict(
-                type='Pretrained', checkpoint='./pretrained/stdc2.pth'))))
+            init_cfg=dict(type='Pretrained', checkpoint=checkpoint))))
diff --git a/configs/twins/README.md b/configs/twins/README.md
@@ -34,9 +34,9 @@ Very recently, a variety of vision transformer architectures for dense predictio
 
 ## Usage
 
-To use other repositories' pre-trained models, it is necessary to convert keys.
+We have provided pretrained models converted from [official repo](https://github.com/Meituan-AutoML/Twins).
 
-We provide a script [`twins2mmseg.py`](../../tools/model_converters/twins2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/Meituan-AutoML/Twins) to MMSegmentation style.
+If you want to convert keys on your own to use official repositories' pre-trained models, we also provide a script [`twins2mmseg.py`](../../tools/model_converters/twins2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/Meituan-AutoML/Twins) to MMSegmentation style.
 
 ```shell
 python tools/model_converters/twins2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH} ${MODEL_TYPE}
diff --git a/configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py b/configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
@@ -1,7 +1,8 @@
 _base_ = ['./twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_base_20220308-0621964c.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_base.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         depths=[3, 4, 18, 3]), )
diff --git a/configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py b/configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py
@@ -1,9 +1,10 @@
 _base_ = ['./twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py']
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_base_20220308-0621964c.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_base.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         depths=[3, 4, 18, 3],
         drop_path_rate=0.3))
 
diff --git a/configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py b/configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
@@ -1,7 +1,8 @@
 _base_ = ['./twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_large_20220308-37579dc6.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_large.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         depths=[3, 8, 27, 3]))
diff --git a/configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py b/configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py
@@ -1,8 +1,10 @@
 _base_ = ['./twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py']
+
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_large_20220308-37579dc6.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/pcpvt_large.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         depths=[3, 8, 27, 3],
         drop_path_rate=0.3))
 
diff --git a/configs/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py b/configs/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
@@ -1,9 +1,10 @@
 _base_ = ['./twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_base_20220308-1b7eb711.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_base.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[96, 192, 384, 768],
         num_heads=[3, 6, 12, 24],
         depths=[2, 2, 18, 2]),
diff --git a/configs/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k.py b/configs/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k.py
@@ -1,8 +1,10 @@
 _base_ = ['./twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py']
+
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_base_20220308-1b7eb711.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_base.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[96, 192, 384, 768],
         num_heads=[3, 6, 12, 24],
         depths=[2, 2, 18, 2]),
diff --git a/configs/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py b/configs/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
@@ -1,9 +1,10 @@
 _base_ = ['./twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
 
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_large_20220308-fb5936f3.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_large.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[128, 256, 512, 1024],
         num_heads=[4, 8, 16, 32],
         depths=[2, 2, 18, 2],
diff --git a/configs/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k.py b/configs/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k.py
@@ -1,8 +1,10 @@
 _base_ = ['./twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py']
+
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_large_20220308-fb5936f3.pth'  # noqa
+
 model = dict(
     backbone=dict(
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_large.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[128, 256, 512, 1024],
         num_heads=[4, 8, 16, 32],
         depths=[2, 2, 18, 2],
diff --git a/configs/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py b/configs/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py
@@ -2,11 +2,13 @@
     '../_base_/models/twins_pcpvt-s_fpn.py', '../_base_/datasets/ade20k.py',
     '../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
 ]
+
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_small_20220308-7e1c3695.pth'  # noqa
+
 model = dict(
     backbone=dict(
         type='SVT',
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_small.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[64, 128, 256, 512],
         num_heads=[2, 4, 8, 16],
         mlp_ratios=[4, 4, 4, 4],
diff --git a/configs/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py b/configs/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py
@@ -3,11 +3,13 @@
     '../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
     '../_base_/schedules/schedule_160k.py'
 ]
+
+checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/alt_gvt_small_20220308-7e1c3695.pth'  # noqa
+
 model = dict(
     backbone=dict(
         type='SVT',
-        init_cfg=dict(
-            type='Pretrained', checkpoint='pretrained/alt_gvt_small.pth'),
+        init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
         embed_dims=[64, 128, 256, 512],
         num_heads=[2, 4, 8, 16],
         mlp_ratios=[4, 4, 4, 4],
