change metafiles (open-mmlab#264)

Research-and-Project · Sep 9, 2021 · eddd85f · eddd85f
1 parent ad084a6
commit eddd85f
Show file tree

Hide file tree

Showing 8 changed files with 52 additions and 27 deletions.
diff --git a/configs/fp16/metafile.yml b/configs/fp16/metafile.yml
@@ -4,19 +4,22 @@ Collections:
       Training Techniques:
         - Mixed Precision Training
       Training Resources: 8x TITAN Xp GPUs
-    Paper: https://arxiv.org/abs/1710.03740
+    Paper:
+        URL: https://arxiv.org/abs/1710.03740
+        Title: Mixed Precision Training
     README: configs/fp16/README.md
 
 Models:
   - Name: selsa_faster_rcnn_r50_dc5_fp16_1x_imagenetvid
     In Collection: FP16
     Config: configs/fp16/selsa_faster_rcnn_r50_dc5_fp16_1x_imagenetvid.py
     Metadata:
+      Training Data: ImageNet DET, ImageNet VID
       Training Memory (GB): 2.71
       Epochs: 7
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 78.7
     Weights: https://download.openmmlab.com/mmtracking/fp16/selsa_faster_rcnn_r50_dc5_fp16_1x_imagenetvid_20210728_193846-dce6eb09.pth
@@ -40,6 +43,8 @@ Models:
     In Collection: FP16
     Config: configs/fp16/siamese_rpn_r50_fp16_1x_lasot.py
     Metadata:
+      Training Data: MSCOCO, ImageNet DET, ImageNet VID
+      Training Memory (GB): 7.54
       Epochs: 20
     Results:
       - Task: Single Object Tracking

diff --git a/configs/mot/deepsort/metafile.yml b/configs/mot/deepsort/metafile.yml
@@ -7,7 +7,9 @@ Collections:
       Architecture:
         - ResNet
         - FPN
-    Paper: https://arxiv.org/abs/1602.00763
+    Paper:
+        URL: https://arxiv.org/abs/1602.00763
+        Title: Simple Online and Realtime Tracking
     README: configs/mot/deepsort/README.md
 
 Models:
@@ -54,7 +56,9 @@ Models:
   - Name: deepsort_faster-rcnn_fpn_4e_mot17-public-half
     In Collection: SORT
     Config: configs/mot/deepsort/deepsort_faster-rcnn_fpn_4e_mot17-public-half.py
-    Paper: https://arxiv.org/abs/1703.07402
+    Paper:
+        URL: https://arxiv.org/abs/1703.07402
+        Title: Simple Online and Realtime Tracking with a Deep Association Metric
     Metadata:
       Training Data: MOT17-half-train
       inference time (ms/im):
@@ -77,7 +81,9 @@ Models:
   - Name: deepsort_faster-rcnn_fpn_4e_mot17-private-half
     In Collection: SORT
     Config: configs/mot/deepsort/deepsort_faster-rcnn_fpn_4e_mot17-private-half.py
-    Paper: https://arxiv.org/abs/1703.07402
+    Paper:
+        URL: https://arxiv.org/abs/1703.07402
+        Title: Simple Online and Realtime Tracking with a Deep Association Metric
     Metadata:
       Training Data: MOT17-half-train
       inference time (ms/im):

diff --git a/configs/mot/tracktor/metafile.yml b/configs/mot/tracktor/metafile.yml
@@ -7,7 +7,9 @@ Collections:
       Architecture:
         - ResNet
         - FPN
-    Paper: https://arxiv.org/abs/1903.05625
+    Paper:
+        URL: https://arxiv.org/abs/1903.05625
+        Title: Tracking without bells and whistles
     README: configs/mot/tracktor/README.md
 
 Models:

diff --git a/configs/sot/siamese_rpn/metafile.yml b/configs/sot/siamese_rpn/metafile.yml
@@ -7,15 +7,19 @@ Collections:
       Training Resources: 8x V100 GPUs
       Architecture:
         - ResNet
-    Paper: https://arxiv.org/abs/1812.11703
+    Paper:
+        URL: https://arxiv.org/abs/1812.11703
+        Title: SiamRPN++ Evolution of Siamese Visual Tracking with Very Deep Networks
     README: configs/sot/siamese_rpn/README.md
 
 Models:
   - Name: siamese_rpn_r50_1x_lasot
     In Collection: SiameseRPN++
     Config: configs/sot/siamese_rpn/siamese_rpn_r50_1x_lasot.py
     Metadata:
+      Training Data: MSCOCO, ImageNet DET, ImageNet VID
       Training Memory (GB): 7.54
+      Epochs: 20
     Results:
       - Task: Single Object Tracking
         Dataset: LaSOT

diff --git a/configs/vid/dff/metafile.yml b/configs/vid/dff/metafile.yml
@@ -1,13 +1,15 @@
 Collections:
   - Name: DFF
     Metadata:
-      Training Data: ILSVRC
+      Training Data: ImageNet DET, ImageNet VID
       Training Techniques:
         - SGD with Momentum
       Training Resources: 8x V100 GPUs
       Architecture:
         - ResNet
-    Paper: https://arxiv.org/abs/1611.07715
+    Paper:
+        URL: https://arxiv.org/abs/1611.07715
+        Title: Deep Feature Flow for Video Recognition
     README: configs/vid/dff/README.md
 
 Models:
@@ -18,7 +20,7 @@ Models:
       Training Memory (GB): 2.50
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 70.3
     Weights: https://download.openmmlab.com/mmtracking/vid/dff/dff_faster_rcnn_r50_dc5_1x_imagenetvid/dff_faster_rcnn_r50_dc5_1x_imagenetvid_20201227_213250-548911a4.pth
@@ -30,7 +32,7 @@ Models:
       Training Memory (GB): 3.25
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 73.5
     Weights: https://download.openmmlab.com/mmtracking/vid/dff/dff_faster_rcnn_r101_dc5_1x_imagenetvid/dff_faster_rcnn_r101_dc5_1x_imagenetvid_20201218_172720-ad732e17.pth
@@ -42,7 +44,7 @@ Models:
       Training Memory (GB): 4.95
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 75.5
     Weights: https://download.openmmlab.com/mmtracking/vid/dff/dff_faster_rcnn_x101_dc5_1x_imagenetvid/dff_faster_rcnn_x101_dc5_1x_imagenetvid_20210819_095932-0a9e6cb5.pth
diff --git a/configs/vid/fgfa/metafile.yml b/configs/vid/fgfa/metafile.yml
@@ -1,13 +1,15 @@
 Collections:
   - Name: FGFA
     Metadata:
-      Training Data: ILSVRC
+      Training Data: ImageNet DET, ImageNet VID
       Training Techniques:
         - SGD with Momentum
       Training Resources: 8x V100 GPUs
       Architecture:
         - ResNet
-    Paper: https://arxiv.org/abs/1703.10025
+    Paper:
+        URL: https://arxiv.org/abs/1703.10025
+        Title: Flow-Guided Feature Aggregation for Video Object Detection
     README: configs/vid/fgfa/README.md
 
 Models:
@@ -18,7 +20,7 @@ Models:
       Training Memory (GB): 4.10
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 74.7
     Weights: https://download.openmmlab.com/mmtracking/vid/fgfa/fgfa_faster_rcnn_r50_dc5_1x_imagenetvid/fgfa_faster_rcnn_r50_dc5_1x_imagenetvid_20201228_022657-f42016f3.pth
@@ -30,7 +32,7 @@ Models:
       Training Memory (GB): 5.80
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 77.8
     Weights: https://download.openmmlab.com/mmtracking/vid/fgfa/fgfa_faster_rcnn_r101_dc5_1x_imagenetvid/fgfa_faster_rcnn_r101_dc5_1x_imagenetvid_20201219_011831-9c9d8183.pth
@@ -42,7 +44,7 @@ Models:
       Training Memory (GB): 9.74
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 79.6
     Weights: https://download.openmmlab.com/mmtracking/vid/fgfa/fgfa_faster_rcnn_x101_dc5_1x_imagenetvid/fgfa_faster_rcnn_x101_dc5_1x_imagenetvid_20210818_223334-8723c594.pth
diff --git a/configs/vid/selsa/metafile.yml b/configs/vid/selsa/metafile.yml
@@ -1,13 +1,15 @@
 Collections:
   - Name: SELSA
     Metadata:
-      Training Data: ILSVRC
+      Training Data: ImageNet DET, ImageNet VID
       Training Techniques:
         - SGD with Momentum
       Training Resources: 8x V100 GPUs
       Architecture:
         - ResNet
-    Paper: https://arxiv.org/abs/1907.06390
+    Paper:
+        URL: https://arxiv.org/abs/1907.06390
+        Title: Sequence Level Semantics Aggregation for Video Object Detection
     README: configs/vid/selsa/README.md
 
 Models:
@@ -18,7 +20,7 @@ Models:
       Training Memory (GB): 3.49
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 78.4
     Weights: https://download.openmmlab.com/mmtracking/vid/selsa/selsa_faster_rcnn_r50_dc5_1x_imagenetvid/selsa_faster_rcnn_r50_dc5_1x_imagenetvid_20201227_204835-2f5a4952.pth
@@ -30,7 +32,7 @@ Models:
       Training Memory (GB): 5.18
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 81.5
     Weights: https://download.openmmlab.com/mmtracking/vid/selsa/selsa_faster_rcnn_r101_dc5_1x_imagenetvid/selsa_faster_rcnn_r101_dc5_1x_imagenetvid_20201218_172724-aa961bcc.pth
@@ -42,7 +44,7 @@ Models:
       Training Memory (GB): 9.15
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 83.1
     Weights: https://download.openmmlab.com/mmtracking/vid/selsa/selsa_faster_rcnn_x101_dc5_1x_imagenetvid/selsa_faster_rcnn_x101_dc5_1x_imagenetvid_20210825_205641-10252965.pth
diff --git a/configs/vid/temporal_roi_align/metafile.yml b/configs/vid/temporal_roi_align/metafile.yml
@@ -1,13 +1,15 @@
 Collections:
   - Name: Temporal RoI Align
     Metadata:
-      Training Data: ILSVRC
+      Training Data: ImageNet DET, ImageNet VID
       Training Techniques:
         - SGD with Momentum
       Training Resources: 8x V100 GPUs
       Architecture:
         - ResNet
-    Paper: https://ojs.aaai.org/index.php/AAAI/article/view/16234
+    Paper:
+        URL: https://ojs.aaai.org/index.php/AAAI/article/view/16234
+        Title: Temporal ROI Align for Video Object Recognition
     README: configs/vid/temporal_roi_align/README.md
 
 Models:
@@ -18,7 +20,7 @@ Models:
       Training Memory (GB): 4.14
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 79.8
     Weights: https://download.openmmlab.com/mmtracking/vid/temporal_roi_align/selsa_troialign_faster_rcnn_r50_dc5_7e_imagenetvid/selsa_troialign_faster_rcnn_r50_dc5_7e_imagenetvid_20210820_162714-939fd657.pth
@@ -30,7 +32,7 @@ Models:
       Training Memory (GB): 5.83
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 82.6
     Weights: https://download.openmmlab.com/mmtracking/vid/temporal_roi_align/selsa_troialign_faster_rcnn_r101_dc5_7e_imagenetvid/selsa_troialign_faster_rcnn_r101_dc5_7e_imagenetvid_20210822_111621-22cb96b9.pth
@@ -42,7 +44,7 @@ Models:
       Training Memory (GB): 9.74
     Results:
       - Task: Video Object Detection
-        Dataset: ILSVRC
+        Dataset: ImageNet VID
         Metrics:
           box [email protected]: 84.1
     Weights: https://download.openmmlab.com/mmtracking/vid/temporal_roi_align/selsa_troialign_faster_rcnn_x101_dc5_7e_imagenetvid/selsa_troialign_faster_rcnn_x101_dc5_7e_imagenetvid_20210822_164036-4471ac42.pth