metafile.yml 8.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263
  1. Collections:
  2. - Name: Weight Standardization
  3. Metadata:
  4. Training Data: COCO
  5. Training Techniques:
  6. - SGD with Momentum
  7. - Weight Decay
  8. Training Resources: 8x V100 GPUs
  9. Architecture:
  10. - Group Normalization
  11. - Weight Standardization
  12. Paper:
  13. URL: https://arxiv.org/abs/1903.10520
  14. Title: 'Weight Standardization'
  15. README: configs/gn+ws/README.md
  16. Code:
  17. URL: https://github.com/open-mmlab/mmdetection/blob/v2.0.0/configs/gn%2Bws/mask-rcnn_r50_fpn_gn-ws-all_2x_coco.py
  18. Version: v2.0.0
  19. Models:
  20. - Name: faster-rcnn_r50_fpn_gn_ws-all_1x_coco
  21. In Collection: Weight Standardization
  22. Config: configs/gn%2Bws/faster-rcnn_r50_fpn_gn-ws-all_1x_coco.py
  23. Metadata:
  24. Training Memory (GB): 5.9
  25. inference time (ms/im):
  26. - value: 85.47
  27. hardware: V100
  28. backend: PyTorch
  29. batch size: 1
  30. mode: FP32
  31. resolution: (800, 1333)
  32. Epochs: 12
  33. Results:
  34. - Task: Object Detection
  35. Dataset: COCO
  36. Metrics:
  37. box AP: 39.7
  38. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/faster_rcnn_r50_fpn_gn_ws-all_1x_coco/faster_rcnn_r50_fpn_gn_ws-all_1x_coco_20200130-613d9fe2.pth
  39. - Name: faster-rcnn_r101_fpn_gn-ws-all_1x_coco
  40. In Collection: Weight Standardization
  41. Config: configs/gn%2Bws/faster-rcnn_r101_fpn_gn-ws-all_1x_coco.py
  42. Metadata:
  43. Training Memory (GB): 8.9
  44. inference time (ms/im):
  45. - value: 111.11
  46. hardware: V100
  47. backend: PyTorch
  48. batch size: 1
  49. mode: FP32
  50. resolution: (800, 1333)
  51. Epochs: 12
  52. Results:
  53. - Task: Object Detection
  54. Dataset: COCO
  55. Metrics:
  56. box AP: 41.7
  57. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/faster_rcnn_r101_fpn_gn_ws-all_1x_coco/faster_rcnn_r101_fpn_gn_ws-all_1x_coco_20200205-a93b0d75.pth
  58. - Name: faster-rcnn_x50-32x4d_fpn_gn-ws-all_1x_coco
  59. In Collection: Weight Standardization
  60. Config: configs/gn%2Bws/faster-rcnn_x50-32x4d_fpn_gn-ws-all_1x_coco.py
  61. Metadata:
  62. Training Memory (GB): 7.0
  63. inference time (ms/im):
  64. - value: 97.09
  65. hardware: V100
  66. backend: PyTorch
  67. batch size: 1
  68. mode: FP32
  69. resolution: (800, 1333)
  70. Epochs: 12
  71. Results:
  72. - Task: Object Detection
  73. Dataset: COCO
  74. Metrics:
  75. box AP: 40.7
  76. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/faster_rcnn_x50_32x4d_fpn_gn_ws-all_1x_coco/faster_rcnn_x50_32x4d_fpn_gn_ws-all_1x_coco_20200203-839c5d9d.pth
  77. - Name: faster-rcnn_x101-32x4d_fpn_gn-ws-all_1x_coco
  78. In Collection: Weight Standardization
  79. Config: configs/gn%2Bws/faster-rcnn_x101-32x4d_fpn_gn-ws-all_1x_coco.py
  80. Metadata:
  81. Training Memory (GB): 10.8
  82. inference time (ms/im):
  83. - value: 131.58
  84. hardware: V100
  85. backend: PyTorch
  86. batch size: 1
  87. mode: FP32
  88. resolution: (800, 1333)
  89. Epochs: 12
  90. Results:
  91. - Task: Object Detection
  92. Dataset: COCO
  93. Metrics:
  94. box AP: 42.1
  95. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/faster_rcnn_x101_32x4d_fpn_gn_ws-all_1x_coco/faster_rcnn_x101_32x4d_fpn_gn_ws-all_1x_coco_20200212-27da1bc2.pth
  96. - Name: mask-rcnn_r50_fpn_gn_ws-all_2x_coco
  97. In Collection: Weight Standardization
  98. Config: configs/gn%2Bws/mask-rcnn_r50_fpn_gn-ws-all_2x_coco.py
  99. Metadata:
  100. Training Memory (GB): 7.3
  101. inference time (ms/im):
  102. - value: 95.24
  103. hardware: V100
  104. backend: PyTorch
  105. batch size: 1
  106. mode: FP32
  107. resolution: (800, 1333)
  108. Epochs: 24
  109. Results:
  110. - Task: Object Detection
  111. Dataset: COCO
  112. Metrics:
  113. box AP: 40.6
  114. - Task: Instance Segmentation
  115. Dataset: COCO
  116. Metrics:
  117. mask AP: 36.6
  118. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_r50_fpn_gn_ws-all_2x_coco/mask_rcnn_r50_fpn_gn_ws-all_2x_coco_20200226-16acb762.pth
  119. - Name: mask-rcnn_r101_fpn_gn-ws-all_2x_coco
  120. In Collection: Weight Standardization
  121. Config: configs/gn%2Bws/mask-rcnn_r101_fpn_gn-ws-all_2x_coco.py
  122. Metadata:
  123. Training Memory (GB): 10.3
  124. inference time (ms/im):
  125. - value: 116.28
  126. hardware: V100
  127. backend: PyTorch
  128. batch size: 1
  129. mode: FP32
  130. resolution: (800, 1333)
  131. Epochs: 24
  132. Results:
  133. - Task: Object Detection
  134. Dataset: COCO
  135. Metrics:
  136. box AP: 42.0
  137. - Task: Instance Segmentation
  138. Dataset: COCO
  139. Metrics:
  140. mask AP: 37.7
  141. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_r101_fpn_gn_ws-all_2x_coco/mask_rcnn_r101_fpn_gn_ws-all_2x_coco_20200212-ea357cd9.pth
  142. - Name: mask-rcnn_x50-32x4d_fpn_gn-ws-all_2x_coco
  143. In Collection: Weight Standardization
  144. Config: configs/gn%2Bws/mask-rcnn_x50-32x4d_fpn_gn-ws-all_2x_coco.py
  145. Metadata:
  146. Training Memory (GB): 8.4
  147. inference time (ms/im):
  148. - value: 107.53
  149. hardware: V100
  150. backend: PyTorch
  151. batch size: 1
  152. mode: FP32
  153. resolution: (800, 1333)
  154. Epochs: 24
  155. Results:
  156. - Task: Object Detection
  157. Dataset: COCO
  158. Metrics:
  159. box AP: 41.1
  160. - Task: Instance Segmentation
  161. Dataset: COCO
  162. Metrics:
  163. mask AP: 37.0
  164. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_x50_32x4d_fpn_gn_ws-all_2x_coco/mask_rcnn_x50_32x4d_fpn_gn_ws-all_2x_coco_20200216-649fdb6f.pth
  165. - Name: mask-rcnn_x101-32x4d_fpn_gn-ws-all_2x_coco
  166. In Collection: Weight Standardization
  167. Config: configs/gn%2Bws/mask-rcnn_x101-32x4d_fpn_gn-ws-all_2x_coco.py
  168. Metadata:
  169. Training Memory (GB): 12.2
  170. inference time (ms/im):
  171. - value: 140.85
  172. hardware: V100
  173. backend: PyTorch
  174. batch size: 1
  175. mode: FP32
  176. resolution: (800, 1333)
  177. Epochs: 24
  178. Results:
  179. - Task: Object Detection
  180. Dataset: COCO
  181. Metrics:
  182. box AP: 42.1
  183. - Task: Instance Segmentation
  184. Dataset: COCO
  185. Metrics:
  186. mask AP: 37.9
  187. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_x101_32x4d_fpn_gn_ws-all_2x_coco/mask_rcnn_x101_32x4d_fpn_gn_ws-all_2x_coco_20200319-33fb95b5.pth
  188. - Name: mask-rcnn_r50_fpn_gn_ws-all_20_23_24e_coco
  189. In Collection: Weight Standardization
  190. Config: configs/gn%2Bws/mask-rcnn_r50_fpn_gn-ws-all_20-23-24e_coco.py
  191. Metadata:
  192. Training Memory (GB): 7.3
  193. Epochs: 24
  194. Results:
  195. - Task: Object Detection
  196. Dataset: COCO
  197. Metrics:
  198. box AP: 41.1
  199. - Task: Instance Segmentation
  200. Dataset: COCO
  201. Metrics:
  202. mask AP: 37.1
  203. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_r50_fpn_gn_ws-all_20_23_24e_coco/mask_rcnn_r50_fpn_gn_ws-all_20_23_24e_coco_20200213-487d1283.pth
  204. - Name: mask-rcnn_r101_fpn_gn-ws-all_20-23-24e_coco
  205. In Collection: Weight Standardization
  206. Config: configs/gn%2Bws/mask-rcnn_r101_fpn_gn-ws-all_20-23-24e_coco.py
  207. Metadata:
  208. Training Memory (GB): 10.3
  209. Epochs: 24
  210. Results:
  211. - Task: Object Detection
  212. Dataset: COCO
  213. Metrics:
  214. box AP: 43.1
  215. - Task: Instance Segmentation
  216. Dataset: COCO
  217. Metrics:
  218. mask AP: 38.6
  219. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_r101_fpn_gn_ws-all_20_23_24e_coco/mask_rcnn_r101_fpn_gn_ws-all_20_23_24e_coco_20200213-57b5a50f.pth
  220. - Name: mask-rcnn_x50-32x4d_fpn_gn-ws-all_20-23-24e_coco
  221. In Collection: Weight Standardization
  222. Config: configs/gn%2Bws/mask-rcnn_x50-32x4d_fpn_gn-ws-all_20-23-24e_coco.py
  223. Metadata:
  224. Training Memory (GB): 8.4
  225. Epochs: 24
  226. Results:
  227. - Task: Object Detection
  228. Dataset: COCO
  229. Metrics:
  230. box AP: 42.1
  231. - Task: Instance Segmentation
  232. Dataset: COCO
  233. Metrics:
  234. mask AP: 38.0
  235. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_x50_32x4d_fpn_gn_ws-all_20_23_24e_coco/mask_rcnn_x50_32x4d_fpn_gn_ws-all_20_23_24e_coco_20200226-969bcb2c.pth
  236. - Name: mask-rcnn_x101-32x4d_fpn_gn-ws-all_20-23-24e_coco
  237. In Collection: Weight Standardization
  238. Config: configs/gn%2Bws/mask-rcnn_x101-32x4d_fpn_gn-ws-all_20-23-24e_coco.py
  239. Metadata:
  240. Training Memory (GB): 12.2
  241. Epochs: 24
  242. Results:
  243. - Task: Object Detection
  244. Dataset: COCO
  245. Metrics:
  246. box AP: 42.7
  247. - Task: Instance Segmentation
  248. Dataset: COCO
  249. Metrics:
  250. mask AP: 38.5
  251. Weights: https://download.openmmlab.com/mmdetection/v2.0/gn%2Bws/mask_rcnn_x101_32x4d_fpn_gn_ws-all_20_23_24e_coco/mask_rcnn_x101_32x4d_fpn_gn_ws-all_20_23_24e_coco_20200316-e6cd35ef.pth