ZIPUnet分割的代码不含有数据集 2.66MB

m0_74055982需要积分:5(1积分=1元)

资源文件列表:

unet-master - 副本.zip 大约有92个文件
  1. unet-master - 副本/.idea/
  2. unet-master - 副本/.idea/.gitignore 176B
  3. unet-master - 副本/.idea/inspectionProfiles/
  4. unet-master - 副本/.idea/inspectionProfiles/profiles_settings.xml 174B
  5. unet-master - 副本/.idea/inspectionProfiles/Project_Default.xml 510B
  6. unet-master - 副本/.idea/misc.xml 272B
  7. unet-master - 副本/.idea/modules.xml 274B
  8. unet-master - 副本/.idea/unet_42-new.iml 472B
  9. unet-master - 副本/.idea/workspace.xml 7.83KB
  10. unet-master - 副本/__pycache__/
  11. unet-master - 副本/bmp转jpg.py 1KB
  12. unet-master - 副本/bmp转png.py 1001B
  13. unet-master - 副本/data/
  14. unet-master - 副本/data/results/
  15. unet-master - 副本/data/Test_Images/
  16. unet-master - 副本/data/Test_Labels/
  17. unet-master - 副本/data/Training_Images/
  18. unet-master - 副本/data/Training_Labels/
  19. unet-master - 副本/images/
  20. unet-master - 副本/images/111/
  21. unet-master - 副本/images/111/ISIC_0000000.jpg 48.79KB
  22. unet-master - 副本/images/111/ISIC_0000000_res.png 4.76KB
  23. unet-master - 副本/images/ISIC_0000000.jpg 48.79KB
  24. unet-master - 副本/images/ISIC_0000000_res.png 4.72KB
  25. unet-master - 副本/images/right.jpeg 26.57KB
  26. unet-master - 副本/images/tmp/
  27. unet-master - 副本/images/tmp/tmp_upload.jpeg 21.27KB
  28. unet-master - 副本/images/UI/
  29. unet-master - 副本/images/UI/logo.jpeg 33.37KB
  30. unet-master - 副本/images/UI/lufei.png 215.7KB
  31. unet-master - 副本/images/UI/right.jpeg 25.45KB
  32. unet-master - 副本/images/UI/up.jpeg 27.88KB
  33. unet-master - 副本/images/up.jpeg 21.27KB
  34. unet-master - 副本/labelme2seg.py 1.02KB
  35. unet-master - 副本/model/
  36. unet-master - 副本/model/__init__.py
  37. unet-master - 副本/model/__pycache__/
  38. unet-master - 副本/model/__pycache__/__init__.cpython-37.pyc 141B
  39. unet-master - 副本/model/__pycache__/__init__.cpython-38.pyc 161B
  40. unet-master - 副本/model/__pycache__/unet_model.cpython-37.pyc 1.34KB
  41. unet-master - 副本/model/__pycache__/unet_model.cpython-38.pyc 1.37KB
  42. unet-master - 副本/model/__pycache__/unet_parts.cpython-37.pyc 2.79KB
  43. unet-master - 副本/model/__pycache__/unet_parts.cpython-38.pyc 2.75KB
  44. unet-master - 副本/model/unet_model.py 1.29KB
  45. unet-master - 副本/model/unet_parts.py 3.39KB
  46. unet-master - 副本/predict.py 1.72KB
  47. unet-master - 副本/requirements.txt 143B
  48. unet-master - 副本/results/
  49. unet-master - 副本/results/confusion_matrix.csv 68B
  50. unet-master - 副本/results/mIoU.png 15.56KB
  51. unet-master - 副本/results/mPA.png 14.85KB
  52. unet-master - 副本/results/Precision.png 14.7KB
  53. unet-master - 副本/results/Recall.png 14.06KB
  54. unet-master - 副本/test.py 4.17KB
  55. unet-master - 副本/testdata/
  56. unet-master - 副本/testdata/jsons/
  57. unet-master - 副本/testdata/jsons/Case-1-U-1-1.json 65.34KB
  58. unet-master - 副本/testdata/jsons/Case-2-U-2-2.json 77.86KB
  59. unet-master - 副本/testdata/jsons/Case-2-U-2-3.json 89.63KB
  60. unet-master - 副本/testdata/jsons/Case-3-U-5-0.json 83.71KB
  61. unet-master - 副本/testdata/jsons/Case-3-U-5-2.json 83.12KB
  62. unet-master - 副本/testdata/jsons/Case-3-U-5-3.json 80.91KB
  63. unet-master - 副本/testdata/jsons/Case-4-U-8-0.json 84.39KB
  64. unet-master - 副本/testdata/jsons/Case-4-U-8-1.json 85.46KB
  65. unet-master - 副本/testdata/jsons/Case-5-U-10-0.json 78.91KB
  66. unet-master - 副本/testdata/jsons/Case-5-U-10-1.json 82.62KB
  67. unet-master - 副本/testdata/labels/
  68. unet-master - 副本/testdata/labels/Case-1-U-1-1.png 3.18KB
  69. unet-master - 副本/testdata/labels/Case-2-U-2-2.png 2.82KB
  70. unet-master - 副本/testdata/labels/Case-2-U-2-3.png 3.25KB
  71. unet-master - 副本/testdata/labels/Case-3-U-5-0.png 3.55KB
  72. unet-master - 副本/testdata/labels/Case-3-U-5-2.png 3.59KB
  73. unet-master - 副本/testdata/labels/Case-3-U-5-3.png 3.24KB
  74. unet-master - 副本/testdata/labels/Case-4-U-8-0.png 3.02KB
  75. unet-master - 副本/testdata/labels/Case-4-U-8-1.png 2.44KB
  76. unet-master - 副本/testdata/labels/Case-5-U-10-0.png 3.47KB
  77. unet-master - 副本/testdata/labels/Case-5-U-10-1.png 3.46KB
  78. unet-master - 副本/train.py 2.46KB
  79. unet-master - 副本/ui.py 8.04KB
  80. unet-master - 副本/unet原文.pdf 1.57MB
  81. unet-master - 副本/utils/
  82. unet-master - 副本/utils/__pycache__/
  83. unet-master - 副本/utils/__pycache__/dataset.cpython-37.pyc 1.81KB
  84. unet-master - 副本/utils/__pycache__/dataset.cpython-38.pyc 1.84KB
  85. unet-master - 副本/utils/__pycache__/utils_metrics.cpython-37.pyc 6.01KB
  86. unet-master - 副本/utils/__pycache__/utils_metrics.cpython-38.pyc 6.35KB
  87. unet-master - 副本/utils/data_remove_seg.py 730B
  88. unet-master - 副本/utils/dataset.py 2.81KB
  89. unet-master - 副本/utils/gen_split.py 1.57KB
  90. unet-master - 副本/utils/label2png.py 1.41KB
  91. unet-master - 副本/utils/utils_metrics.py 9.26KB
  92. unet-master - 副本/切换镜像.txt 348B

资源介绍:

Unet分割的代码不含有数据集
<link href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/css/base.min.css" rel="stylesheet"/><link href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/css/fancy.min.css" rel="stylesheet"/><link href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/89912414/raw.css" rel="stylesheet"/><div id="sidebar" style="display: none"><div id="outline"></div></div><div class="pf w0 h0" data-page-no="1" id="pf1"><div class="pc pc1 w0 h0"><img alt="" class="bi x0 y0 w1 h1" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/89912414/bg1.jpg"/><div class="t m0 x1 h2 y1 ff1 fs0 fc0 sc0 ls0 ws0">U-Net:<span class="_ _0"> </span>Con<span class="_ _1"></span>v<span class="_ _1"></span>olutional<span class="_ _0"> </span>Net<span class="_ _1"></span>w<span class="_ _1"></span>orks<span class="_ _0"> </span>for<span class="_ _0"> </span>Biomedical</div><div class="t m0 x2 h2 y2 ff1 fs0 fc0 sc0 ls0 ws0">Image<span class="_ _0"> </span>Segmen<span class="_ _1"></span>tation</div><div class="t m0 x3 h3 y3 ff2 fs1 fc0 sc0 ls0 ws0">Olaf<span class="_ _2"> </span>Ronneb<span class="_ _3"></span>erger,<span class="_ _2"> </span>Philipp<span class="_ _2"> </span>Fisc<span class="_ _1"></span>her,<span class="_ _2"> </span>and<span class="_ _2"> </span>Thomas<span class="_ _2"> </span>Brox</div><div class="t m0 x4 h4 y4 ff3 fs2 fc0 sc0 ls0 ws0">Computer<span class="_ _4"> </span>Science<span class="_ _4"> </span>Departmen<span class="_ _1"></span>t<span class="_ _4"> </span>and<span class="_ _4"> </span>BIOSS<span class="_ _4"> </span>Centre<span class="_ _4"> </span>for<span class="_ _4"> </span>Biological<span class="_ _4"> </span>Signalling<span class="_ _4"> </span>Studies,</div><div class="t m0 x5 h4 y5 ff3 fs2 fc0 sc0 ls0 ws0">Univ<span class="_ _1"></span>ersity<span class="_ _4"> </span>of<span class="_ _4"> </span>F<span class="_ _5"></span>reiburg,<span class="_ _4"> </span>German<span class="_ _1"></span>y</div><div class="t m0 x6 h4 y6 ff4 fs2 fc0 sc0 ls0 ws0">ronneber@informatik.uni-freiburg.de<span class="ff3">,</span></div><div class="t m0 x7 h4 y7 ff3 fs2 fc0 sc0 ls0 ws0">WWW<span class="_ _4"> </span>home<span class="_ _4"> </span>page:<span class="_ _4"> </span><span class="ff4">http://lmb.informatik.uni-freiburg.de/</span></div><div class="t m0 x8 h4 y8 ff5 fs2 fc0 sc0 ls0 ws0">Abstract.<span class="_ _6"> </span><span class="ff3">There<span class="_ _7"> </span>is<span class="_ _7"> </span>large<span class="_ _7"> </span>consent<span class="_ _2"> </span>that<span class="_ _7"> </span>successful<span class="_ _7"> </span>training<span class="_ _7"> </span>of<span class="_ _7"> </span>deep<span class="_ _7"> </span>net-</span></div><div class="t m0 x8 h4 y9 ff3 fs2 fc0 sc0 ls0 ws0">w<span class="_ _1"></span>orks<span class="_ _8"> </span>requires<span class="_ _7"> </span>many<span class="_ _2"> </span>thousand<span class="_ _8"> </span>annotated<span class="_ _7"> </span>training<span class="_ _8"> </span>samples.<span class="_ _7"> </span>In<span class="_ _7"> </span>this<span class="_ _8"> </span>pa-</div><div class="t m0 x8 h4 ya ff3 fs2 fc0 sc0 ls0 ws0">p<span class="_ _3"></span>er,<span class="_"> </span>we<span class="_"> </span>present<span class="_"> </span>a<span class="_"> </span>netw<span class="_ _1"></span>ork<span class="_ _4"> </span>and<span class="_"> </span>training<span class="_"> </span>strategy<span class="_ _4"> </span>that<span class="_"> </span>relies<span class="_ _4"> </span>on<span class="_"> </span>the<span class="_"> </span>strong</div><div class="t m0 x8 h4 yb ff3 fs2 fc0 sc0 ls0 ws0">use<span class="_ _2"> </span>of<span class="_ _7"> </span>data<span class="_ _7"> </span>augmentation<span class="_ _2"> </span>to<span class="_ _7"> </span>use<span class="_ _7"> </span>the<span class="_ _7"> </span>av<span class="_ _5"></span>ailable<span class="_ _7"> </span>annotated<span class="_ _7"> </span>samples<span class="_ _7"> </span>more</div><div class="t m0 x8 h4 yc ff3 fs2 fc0 sc0 ls0 ws0">efficien<span class="_ _1"></span>tly<span class="_ _1"></span>.<span class="_ _9"> </span>The<span class="_ _9"> </span>architecture<span class="_ _9"> </span>consists<span class="_ _9"> </span>of<span class="_ _a"> </span>a<span class="_ _9"> </span>contracting<span class="_ _9"> </span>path<span class="_ _9"> </span>to<span class="_ _a"> </span>capture</div><div class="t m0 x8 h4 yd ff3 fs2 fc0 sc0 ls0 ws0">con<span class="_ _1"></span>text<span class="_ _7"> </span>and<span class="_ _7"> </span>a<span class="_ _7"> </span>symmetric<span class="_ _7"> </span>expanding<span class="_ _7"> </span>path<span class="_ _7"> </span>that<span class="_ _7"> </span>enables<span class="_ _7"> </span>precise<span class="_ _7"> </span>lo<span class="_ _3"></span>caliza-</div><div class="t m0 x8 h4 ye ff3 fs2 fc0 sc0 ls0 ws0">tion.<span class="_ _4"> </span>W<span class="_ _5"></span>e<span class="_ _4"> </span>sho<span class="_ _1"></span>w<span class="_ _4"> </span>that<span class="_ _4"> </span>suc<span class="_ _1"></span>h<span class="_ _4"> </span>a<span class="_ _4"> </span>netw<span class="_ _1"></span>ork<span class="_ _4"> </span>can<span class="_ _4"> </span>be<span class="_ _4"> </span>trained<span class="_ _4"> </span>end-to-end<span class="_ _4"> </span>from<span class="_ _4"> </span>v<span class="_ _1"></span>ery</div><div class="t m0 x8 h4 yf ff3 fs2 fc0 sc0 ls0 ws0">few<span class="_ _9"> </span>images<span class="_ _8"> </span>and<span class="_ _9"> </span>outperforms<span class="_ _9"> </span>the<span class="_ _9"> </span>prior<span class="_ _8"> </span>b<span class="_ _3"></span>est<span class="_ _9"> </span>metho<span class="_ _3"></span>d<span class="_ _9"> </span>(a<span class="_ _8"> </span>sliding-window</div><div class="t m0 x8 h4 y10 ff3 fs2 fc0 sc0 ls0 ws0">con<span class="_ _1"></span>volutional<span class="_ _7"> </span>netw<span class="_ _1"></span>ork)<span class="_ _7"> </span>on<span class="_ _8"> </span>the<span class="_ _7"> </span>ISBI<span class="_ _8"> </span>challenge<span class="_ _7"> </span>for<span class="_ _7"> </span>segmentation<span class="_ _7"> </span>of<span class="_ _8"> </span>neu-</div><div class="t m0 x8 h4 y11 ff3 fs2 fc0 sc0 ls0 ws0">ronal<span class="_ _6"> </span>structures<span class="_ _a"> </span>in<span class="_ _6"> </span>electron<span class="_ _6"> </span>microscopic<span class="_ _6"> </span>stacks.<span class="_ _a"> </span>Using<span class="_ _6"> </span>the<span class="_ _6"> </span>same<span class="_ _a"> </span>net-</div><div class="t m0 x8 h4 y12 ff3 fs2 fc0 sc0 ls0 ws0">w<span class="_ _1"></span>ork<span class="_ _a"> </span>trained<span class="_ _9"> </span>on<span class="_ _9"> </span>transmitted<span class="_ _a"> </span>light<span class="_ _9"> </span>microscop<span class="_ _1"></span>y<span class="_ _9"> </span>images<span class="_ _a"> </span>(phase<span class="_ _9"> </span>contrast</div><div class="t m0 x8 h4 y13 ff3 fs2 fc0 sc0 ls0 ws0">and<span class="_ _8"> </span>DIC)<span class="_ _8"> </span>we<span class="_ _7"> </span>won<span class="_ _8"> </span>the<span class="_ _8"> </span>ISBI<span class="_ _8"> </span>cell<span class="_ _8"> </span>tracking<span class="_ _7"> </span>challenge<span class="_ _8"> </span>2015<span class="_ _8"> </span>in<span class="_ _8"> </span>these<span class="_ _8"> </span>cate-</div><div class="t m0 x8 h4 y14 ff3 fs2 fc0 sc0 ls0 ws0">gories<span class="_ _8"> </span>b<span class="_ _1"></span>y<span class="_ _8"> </span>a<span class="_ _8"> </span>large<span class="_ _8"> </span>margin.<span class="_ _8"> </span>Moreov<span class="_ _1"></span>er,<span class="_ _8"> </span>the<span class="_ _8"> </span>netw<span class="_ _1"></span>ork<span class="_ _8"> </span>is<span class="_ _7"> </span>fast.<span class="_ _8"> </span>Segmentation</div><div class="t m0 x8 h4 y15 ff3 fs2 fc0 sc0 ls0 ws0">of<span class="_ _4"> </span>a<span class="_ _2"> </span>512x512<span class="_ _2"> </span>image<span class="_ _2"> </span>takes<span class="_ _4"> </span>less<span class="_ _2"> </span>than<span class="_ _2"> </span>a<span class="_ _2"> </span>second<span class="_ _2"> </span>on<span class="_ _2"> </span>a<span class="_ _2"> </span>recent<span class="_ _4"> </span>GPU.<span class="_ _2"> </span>The<span class="_ _2"> </span>full</div><div class="t m0 x8 h4 y16 ff3 fs2 fc0 sc0 ls0 ws0">implemen<span class="_ _1"></span>tation<span class="_ _4"> </span>(based<span class="_ _4"> </span>on<span class="_ _4"> </span>Caffe)<span class="_ _4"> </span>and<span class="_ _4"> </span>the<span class="_ _4"> </span>trained<span class="_ _4"> </span>net<span class="_ _1"></span>w<span class="_ _1"></span>orks<span class="_ _4"> </span>are<span class="_ _4"> </span>av<span class="_ _5"></span>ailable</div><div class="t m0 x8 h4 y17 ff3 fs2 fc0 sc0 ls0 ws0">at<span class="_ _4"> </span><span class="fc1">h<span class="_ _1"></span>ttp://lmb.informatik.uni-freiburg.de/people/ronneb<span class="_ _3"></span>er/u-net<span class="fc0">.</span></span></div><div class="t m0 x9 h5 y18 ff1 fs3 fc0 sc0 ls0 ws0">1<span class="_ _b"> </span>In<span class="_ _1"></span>tro<span class="_ _3"></span>duction</div><div class="t m0 x9 h3 y19 ff2 fs1 fc0 sc0 ls0 ws0">In<span class="_ _c"> </span>the<span class="_ _c"> </span>last<span class="_ _c"> </span>t<span class="_ _1"></span>wo<span class="_ _c"> </span>y<span class="_ _1"></span>ears,<span class="_ _c"> </span>deep<span class="_ _c"> </span>conv<span class="_ _1"></span>olutional<span class="_ _c"> </span>net<span class="_ _1"></span>works<span class="_ _c"> </span>ha<span class="_ _1"></span>v<span class="_ _1"></span>e<span class="_"> </span>outperformed<span class="_"> </span>the<span class="_ _c"> </span>state<span class="_ _c"> </span>of</div><div class="t m0 x9 h3 y1a ff2 fs1 fc0 sc0 ls0 ws0">the<span class="_"> </span>art<span class="_"> </span>in<span class="_"> </span>many<span class="_"> </span>visual<span class="_"> </span>recognition<span class="_"> </span>tasks,<span class="_"> </span>e.g.<span class="_"> </span>[<span class="fc2">7</span>,<span class="fc2">3</span>].<span class="_"> </span>While<span class="_"> </span>conv<span class="_ _1"></span>olutional<span class="_"> </span>netw<span class="_ _1"></span>orks</div><div class="t m0 x9 h3 y1b ff2 fs1 fc0 sc0 ls0 ws0">ha<span class="_ _1"></span>ve<span class="_ _9"> </span>already<span class="_ _9"> </span>existed<span class="_ _9"> </span>for<span class="_ _9"> </span>a<span class="_ _9"> </span>long<span class="_ _9"> </span>time<span class="_ _9"> </span>[<span class="fc2">8</span>],<span class="_ _9"> </span>their<span class="_ _9"> </span>success<span class="_ _9"> </span>was<span class="_ _9"> </span>limited<span class="_ _9"> </span>due<span class="_ _9"> </span>to<span class="_ _9"> </span>the</div><div class="t m0 x9 h3 y1c ff2 fs1 fc0 sc0 ls0 ws0">size<span class="_ _7"> </span>of<span class="_ _8"> </span>the<span class="_ _8"> </span>a<span class="_ _1"></span>v<span class="_ _1"></span>ailable<span class="_ _7"> </span>training<span class="_ _8"> </span>sets<span class="_ _8"> </span>and<span class="_ _7"> </span>the<span class="_ _8"> </span>size<span class="_ _8"> </span>of<span class="_ _7"> </span>the<span class="_ _8"> </span>considered<span class="_ _8"> </span>net<span class="_ _1"></span>works.<span class="_ _7"> </span>The</div><div class="t m0 x9 h3 y1d ff2 fs1 fc0 sc0 ls0 ws0">breakthrough<span class="_ _2"> </span>by<span class="_ _4"> </span>Krizhevsky<span class="_ _7"> </span>et<span class="_ _7"> </span>al.<span class="_ _2"> </span>[<span class="fc2">7</span>]<span class="_ _7"> </span>was<span class="_ _4"> </span>due<span class="_ _7"> </span>to<span class="_ _2"> </span>sup<span class="_ _3"></span>ervised<span class="_ _7"> </span>training<span class="_ _2"> </span>of<span class="_ _7"> </span>a<span class="_ _2"> </span>large</div><div class="t m0 x9 h3 y1e ff2 fs1 fc0 sc0 ls0 ws0">net<span class="_ _1"></span>work<span class="_ _4"> </span>with<span class="_ _4"> </span>8<span class="_ _2"> </span>lay<span class="_ _1"></span>ers<span class="_ _4"> </span>and<span class="_ _2"> </span>millions<span class="_ _4"> </span>of<span class="_ _2"> </span>parameters<span class="_ _4"> </span>on<span class="_ _2"> </span>the<span class="_ _2"> </span>ImageNet<span class="_ _4"> </span>dataset<span class="_ _2"> </span>with</div><div class="t m0 x9 h3 y1f ff2 fs1 fc0 sc0 ls0 ws0">1<span class="_"> </span>million<span class="_"> </span>training<span class="_"> </span>images.<span class="_ _c"> </span>Since<span class="_"> </span>then,<span class="_"> </span>ev<span class="_ _1"></span>en<span class="_"> </span>larger<span class="_"> </span>and<span class="_"> </span>deeper<span class="_"> </span>net<span class="_ _1"></span>works<span class="_"> </span>ha<span class="_ _1"></span>v<span class="_ _1"></span>e<span class="_"> </span>b<span class="_ _3"></span>een</div><div class="t m0 x9 h3 y20 ff2 fs1 fc0 sc0 ls0 ws0">trained<span class="_ _2"> </span>[<span class="fc2">12</span>].</div><div class="t m0 xa h3 y21 ff2 fs1 fc0 sc0 ls0 ws0">The<span class="_ _9"> </span>typical<span class="_ _9"> </span>use<span class="_ _9"> </span>of<span class="_ _9"> </span>con<span class="_ _1"></span>volutional<span class="_ _9"> </span>net<span class="_ _1"></span>works<span class="_ _9"> </span>is<span class="_ _9"> </span>on<span class="_ _9"> </span>classification<span class="_ _9"> </span>tasks,<span class="_ _9"> </span>where</div><div class="t m0 x9 h3 y22 ff2 fs1 fc0 sc0 ls0 ws0">the<span class="_ _8"> </span>output<span class="_ _9"> </span>to<span class="_ _8"> </span>an<span class="_ _8"> </span>image<span class="_ _9"> </span>is<span class="_ _8"> </span>a<span class="_ _8"> </span>single<span class="_ _9"> </span>class<span class="_ _8"> </span>lab<span class="_ _3"></span>el.<span class="_ _8"> </span>How<span class="_ _1"></span>ever,<span class="_ _8"> </span>in<span class="_ _8"> </span>many<span class="_ _8"> </span>visual<span class="_ _8"> </span>tasks,</div><div class="t m0 x9 h3 y23 ff2 fs1 fc0 sc0 ls0 ws0">esp<span class="_ _3"></span>ecially<span class="_ _a"> </span>in<span class="_ _6"> </span>biomedical<span class="_ _6"> </span>image<span class="_ _a"> </span>pro<span class="_ _3"></span>cessing,<span class="_ _6"> </span>the<span class="_ _6"> </span>desired<span class="_ _a"> </span>output<span class="_ _6"> </span>should<span class="_ _6"> </span>include</div><div class="t m0 x9 h3 y24 ff2 fs1 fc0 sc0 ls0 ws0">lo<span class="_ _3"></span>calization,<span class="_ _8"> </span>i.e.,<span class="_ _9"> </span>a<span class="_ _8"> </span>class<span class="_ _9"> </span>lab<span class="_ _3"></span>el<span class="_ _9"> </span>is<span class="_ _8"> </span>supp<span class="_ _3"></span>osed<span class="_ _9"> </span>to<span class="_ _8"> </span>b<span class="_ _3"></span>e<span class="_ _9"> </span>assigned<span class="_ _8"> </span>to<span class="_ _9"> </span>eac<span class="_ _1"></span>h<span class="_ _9"> </span>pixel.<span class="_ _8"> </span>More-</div><div class="t m0 x9 h3 y25 ff2 fs1 fc0 sc0 ls0 ws0">o<span class="_ _1"></span>ver,<span class="_"> </span>thousands<span class="_"> </span>of<span class="_"> </span>training<span class="_ _4"> </span>images<span class="_"> </span>are<span class="_"> </span>usually<span class="_ _4"> </span>b<span class="_ _3"></span>ey<span class="_ _1"></span>ond<span class="_ _4"> </span>reac<span class="_ _1"></span>h<span class="_"> </span>in<span class="_ _4"> </span>biomedical<span class="_"> </span>tasks.</div><div class="t m0 x9 h3 y26 ff2 fs1 fc0 sc0 ls0 ws0">Hence,<span class="_ _2"> </span>Ciresan<span class="_ _2"> </span>et<span class="_ _2"> </span>al.<span class="_ _2"> </span>[<span class="fc2">1</span>]<span class="_ _2"> </span>trained<span class="_ _2"> </span>a<span class="_ _7"> </span>net<span class="_ _1"></span>work<span class="_ _4"> </span>in<span class="_ _2"> </span>a<span class="_ _7"> </span>sliding-windo<span class="_ _1"></span>w<span class="_ _2"> </span>setup<span class="_ _7"> </span>to<span class="_ _2"> </span>predict</div><div class="t m0 x9 h3 y27 ff2 fs1 fc0 sc0 ls0 ws0">the<span class="_"> </span>class<span class="_"> </span>lab<span class="_ _3"></span>el<span class="_"> </span>of<span class="_"> </span>eac<span class="_ _1"></span>h<span class="_"> </span>pixel<span class="_"> </span>by<span class="_"> </span>pro<span class="_ _1"></span>viding<span class="_"> </span>a<span class="_"> </span>lo<span class="_ _3"></span>cal<span class="_"> </span>region<span class="_"> </span>(patch)<span class="_"> </span>around<span class="_"> </span>that<span class="_"> </span>pixel</div><div class="t m1 xb h6 y28 ff6 fs4 fc3 sc0 ls0 ws0">arXiv:1505.04597v1 [cs.CV] 18 May 2015</div><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a></div><div class="pi" data-data='{"ctm":[1.568627,0.000000,0.000000,1.568627,0.000000,0.000000]}'></div></div><div id="pf2" class="pf w0 h0" data-page-no="2"><div class="pc pc2 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/89912414/bg2.jpg"><div class="t m0 x9 h4 y29 ff3 fs2 fc0 sc0 ls0 ws0">2</div><div class="c x9 y2a w2 h7"><div class="t m0 xc h8 y2b ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">co</span><span class="fc9 sc0">p</span><span class="fc9 sc0">y</span><span class="_ _1"></span><span class="fc9 sc0"> </span><span class="fc9 sc0">a</span><span class="fc9 sc0">n</span><span class="fc9 sc0">d</span><span class="fc9 sc0"> </span><span class="fc9 sc0">c</span><span class="fc9 sc0">r</span><span class="fc9 sc0">o</span><span class="fc9 sc0">p</span></div><div class="t m0 xd h8 y2c ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">i</span><span class="fc9 sc0">n</span><span class="fc9 sc0">p</span><span class="fc9 sc0">u</span><span class="fc9 sc0">t</span></div><div class="t m0 xe h8 y2d ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">i</span><span class="fc9 sc0">m</span><span class="fc9 sc0">a</span><span class="fc9 sc0">g</span><span class="fc9 sc0">e</span></div><div class="t m0 xf h8 y2e ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">t</span><span class="fc9 sc0">i</span><span class="fc9 sc0">l</span><span class="fc9 sc0">e</span></div><div class="t m0 x10 h8 y2f ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">o</span><span class="fc9 sc0">u</span><span class="fc9 sc0">t</span><span class="fc9 sc0">p</span><span class="fc9 sc0">u</span><span class="fc9 sc0">t</span><span class="fc9 sc0"> </span></div><div class="t m0 x10 h8 y30 ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">se</span><span class="fc9 sc0">g</span><span class="fc9 sc0">m</span><span class="fc9 sc0">e</span><span class="fc9 sc0">n</span><span class="fc9 sc0">t</span><span class="fc9 sc0">a</span><span class="fc9 sc0">t</span><span class="fc9 sc0">i</span><span class="fc9 sc0">o</span><span class="fc9 sc0">n</span><span class="fc9 sc0"> </span></div><div class="t m0 x10 h8 y31 ff7 fs5 fc0 sc0 ls0 ws0"><span class="fc9 sc0">m</span><span class="fc9 sc0">a</span><span class="fc9 sc0">p</span></div><div class="t m0 x11 h9 y32 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">4</span><span class="_ _d"></span><span class="fc9 sc0">1</span></div><div class="t m0 x12 h9 y33 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">28</span></div><div class="t m0 x13 h9 y34 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">56</span></div><div class="t m0 x14 h9 y35 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">1</span><span class="fc9 sc0">2</span></div><div class="t m0 x1 h9 y36 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">10</span><span class="fc9 sc0">2</span><span class="fc9 sc0">4</span></div><div class="t m0 xc h8 y37 ff7 fs5 fc5 sc0 ls0 ws0"><span class="fc9 sc0">m</span><span class="fc9 sc0">a</span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="fc9 sc0">p</span><span class="fc9 sc0">o</span><span class="fc9 sc0">o</span><span class="fc9 sc0">l</span><span class="fc9 sc0"> </span><span class="fc9 sc0">2</span><span class="fc9 sc0">x</span><span class="fc9 sc0">2</span></div><div class="t m0 x15 h8 y38 ff7 fs5 fc6 sc0 ls0 ws0"><span class="fc9 sc0">u</span><span class="fc9 sc0">p</span><span class="fc9 sc0">-</span><span class="fc9 sc0">c</span><span class="fc9 sc0">o</span><span class="fc9 sc0">n</span><span class="fc9 sc0">v </span><span class="fc9 sc0">2</span><span class="fc9 sc0">x</span><span class="fc9 sc0">2</span></div><div class="t m0 xc h8 y39 ff7 fs5 fc7 sc0 ls0 ws0"><span class="fc9 sc0">c</span><span class="fc9 sc0">o</span><span class="fc9 sc0">n</span><span class="fc9 sc0">v</span><span class="fc9 sc0"> </span><span class="fc9 sc0">3</span><span class="fc9 sc0">x3</span><span class="fc9 sc0">,</span><span class="fc9 sc0"> </span><span class="fc9 sc0">R</span><span class="fc9 sc0">e</span><span class="fc9 sc0">L</span><span class="fc9 sc0">U</span></div><div class="t m1 x16 h9 y3a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">7</span><span class="fc9 sc0">2</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">5</span><span class="fc9 sc0">7</span><span class="fc9 sc0">2</span></div><div class="t m1 x17 h9 y3b ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">8</span><span class="fc9 sc0">4</span><span class="fc9 sc0">&#178;</span></div><div class="t m0 x17 h9 y32 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">4</span></div><div class="t m0 x18 h9 y33 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">28</span></div><div class="t m0 x19 h9 y34 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">56</span></div><div class="t m0 x1a h9 y35 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">12</span></div><div class="t m1 x1b h9 y3a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">7</span><span class="fc9 sc0">0</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">5</span><span class="fc9 sc0">7</span><span class="fc9 sc0">0</span></div><div class="t m1 x17 h9 y3a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">6</span><span class="fc9 sc0">8</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">5</span><span class="fc9 sc0">6</span><span class="fc9 sc0">8</span></div><div class="t m1 x1c h9 y3b ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">8</span><span class="fc9 sc0">2</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x1d h9 y3b ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">8</span><span class="fc9 sc0">0</span><span class="fc9 sc0">&#178;</span><span class="_ _e"></span><span class="fc9 sc0">1</span><span class="fc9 sc0">4</span><span class="fc9 sc0">0</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x1e h9 y3c ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">3</span><span class="fc9 sc0">8</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x1f h9 y3c ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">3</span><span class="fc9 sc0">6</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x1f h9 y3d ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">8</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x20 h9 y3e ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">6</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x21 h9 y3f ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">4</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x21 h9 y40 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">2</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x22 h9 y41 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">8</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x23 h9 y42 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">6</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x24 h9 y43 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">4</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x25 h9 y43 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">2</span><span class="fc9 sc0">&#178;</span></div><div class="t m0 x3 h9 y44 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">51</span><span class="fc9 sc0">2</span></div><div class="t m1 x26 h9 y45 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">0</span><span class="fc9 sc0">4</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x27 h9 y46 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">0</span><span class="fc9 sc0">2</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x2 h9 y46 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">0</span><span class="fc9 sc0">0</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x28 h9 y47 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">0</span><span class="fc9 sc0">0</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x29 h9 y41 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">0</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x2a h9 y48 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">9</span><span class="fc9 sc0">8</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x2b h9 y48 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">9</span><span class="fc9 sc0">6</span><span class="fc9 sc0">&#178;</span></div><div class="t m1 x2c h9 y49 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">9</span><span class="fc9 sc0">2</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">3</span><span class="fc9 sc0">9</span><span class="fc9 sc0">2</span></div><div class="t m1 x2d h9 y4a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">9</span><span class="fc9 sc0">0</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">3</span><span class="fc9 sc0">9</span><span class="fc9 sc0">0</span></div><div class="t m1 x2e h9 y4a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">8</span><span class="fc9 sc0">8</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">3</span><span class="fc9 sc0">8</span><span class="fc9 sc0">8</span></div><div class="t m1 x2f h9 y4a ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">3</span><span class="fc9 sc0">8</span><span class="fc9 sc0">8</span><span class="fc9 sc0"> </span><span class="fc9 sc0">x</span><span class="fc9 sc0"> </span><span class="_ _1"></span><span class="fc9 sc0">3</span><span class="fc9 sc0">8</span><span class="fc9 sc0">8</span></div><div class="t m0 x22 h9 y4b ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">10</span><span class="fc9 sc0">2</span><span class="fc9 sc0">4</span></div><div class="t m0 x25 h9 y4c ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">5</span><span class="fc9 sc0">12</span></div><div class="t m0 x27 h9 y4d ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">5</span><span class="fc9 sc0">6</span></div><div class="t m0 x28 h9 y4e ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">2</span><span class="fc9 sc0">56</span></div><div class="t m0 x30 h9 y4f ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">28</span></div><div class="t m0 x31 h9 y50 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">4</span></div><div class="t m0 x32 h9 y51 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">1</span><span class="fc9 sc0">2</span><span class="fc9 sc0">8</span></div><div class="t m0 x33 h9 y52 ff7 fs6 fc4 sc0 ls0 ws0"><span class="fc9 sc0">6</span><span class="fc9 sc0">4</span><span class="_ _a"> </span><span class="fc9 sc0">2</span></div><div class="t m0 x15 h8 y53 ff7 fs5 fc8 sc0 ls0 ws0"><span class="fc9 sc0">c</span><span class="fc9 sc0">o</span><span class="fc9 sc0">n</span><span class="fc9 sc0">v</span><span class="fc9 sc0"> </span><span class="fc9 sc0">1</span><span class="fc9 sc0">x1</span></div></div><div class="t m0 x9 h4 y54 ff5 fs2 fc0 sc0 ls0 ws0">Fig.<span class="_ _f"> </span>1.<span class="_ _10"> </span><span class="ff3">U-net<span class="_ _10"> </span>arc<span class="_ _1"></span>hitecture<span class="_"> </span>(example<span class="_ _f"> </span>for<span class="_"> </span>32x32<span class="_ _f"> </span>pixels<span class="_"> </span>in<span class="_ _f"> </span>the<span class="_"> </span>lo<span class="_ _1"></span>w<span class="_ _1"></span>est<span class="_"> </span>resolution).<span class="_ _f"> </span>Each<span class="_ _10"> </span>blue</span></div><div class="t m0 x9 h4 y55 ff3 fs2 fc0 sc0 ls0 ws0">b<span class="_ _3"></span>o<span class="_ _1"></span>x<span class="_ _7"> </span>corresp<span class="_ _3"></span>onds<span class="_ _7"> </span>to<span class="_ _7"> </span>a<span class="_ _7"> </span>multi-c<span class="_ _1"></span>hannel<span class="_ _2"> </span>feature<span class="_ _7"> </span>map.<span class="_ _7"> </span>The<span class="_ _7"> </span>num<span class="_ _1"></span>ber<span class="_ _7"> </span>of<span class="_ _7"> </span>channels<span class="_ _2"> </span>is<span class="_ _7"> </span>denoted</div><div class="t m0 x9 h4 y56 ff3 fs2 fc0 sc0 ls0 ws0">on<span class="_ _7"> </span>top<span class="_ _8"> </span>of<span class="_ _7"> </span>the<span class="_ _7"> </span>b<span class="_ _3"></span>ox.<span class="_ _7"> </span>The<span class="_ _7"> </span>x-y-size<span class="_ _8"> </span>is<span class="_ _7"> </span>provided<span class="_ _7"> </span>at<span class="_ _7"> </span>the<span class="_ _8"> </span>low<span class="_ _1"></span>er<span class="_ _7"> </span>left<span class="_ _7"> </span>edge<span class="_ _8"> </span>of<span class="_ _7"> </span>the<span class="_ _8"> </span>b<span class="_ _3"></span>ox.<span class="_ _2"> </span>White</div><div class="t m0 x9 h4 y57 ff3 fs2 fc0 sc0 ls0 ws0">b<span class="_ _3"></span>o<span class="_ _1"></span>xes<span class="_ _4"> </span>represent<span class="_ _4"> </span>copied<span class="_ _4"> </span>feature<span class="_ _4"> </span>maps.<span class="_ _4"> </span>The<span class="_ _4"> </span>arro<span class="_ _1"></span>ws<span class="_ _4"> </span>denote<span class="_ _4"> </span>the<span class="_ _4"> </span>di&#64256;erent<span class="_ _4"> </span>operations.</div><div class="t m0 x9 h3 y58 ff2 fs1 fc0 sc0 ls0 ws0">as<span class="_ _8"> </span>input.<span class="_ _7"> </span>First,<span class="_ _8"> </span>this<span class="_ _8"> </span>netw<span class="_ _1"></span>ork<span class="_ _8"> </span>can<span class="_ _8"> </span>lo<span class="_ _3"></span>calize.<span class="_ _8"> </span>Secondly<span class="_ _5"></span>,<span class="_ _7"> </span>the<span class="_ _8"> </span>training<span class="_ _8"> </span>data<span class="_ _8"> </span>in<span class="_ _8"> </span>terms</div><div class="t m0 x9 h3 y59 ff2 fs1 fc0 sc0 ls0 ws0">of<span class="_ _9"> </span>patches<span class="_ _9"> </span>is<span class="_ _a"> </span>muc<span class="_ _1"></span>h<span class="_ _9"> </span>larger<span class="_ _a"> </span>than<span class="_ _a"> </span>the<span class="_ _9"> </span>num<span class="_ _1"></span>b<span class="_ _3"></span>er<span class="_ _a"> </span>of<span class="_ _a"> </span>training<span class="_ _9"> </span>images.<span class="_ _a"> </span>The<span class="_ _a"> </span>resulting</div><div class="t m0 x9 h3 y5a ff2 fs1 fc0 sc0 ls0 ws0">net<span class="_ _1"></span>work<span class="_ _4"> </span>won<span class="_ _4"> </span>the<span class="_ _2"> </span>EM<span class="_ _2"> </span>segmentation<span class="_ _4"> </span>challenge<span class="_ _4"> </span>at<span class="_ _2"> </span>ISBI<span class="_ _2"> </span>2012<span class="_ _2"> </span>by<span class="_ _4"> </span>a<span class="_ _2"> </span>large<span class="_ _2"> </span>margin.</div><div class="t m0 xa h3 y5b ff2 fs1 fc0 sc0 ls0 ws0">Ob<span class="_ _1"></span>viously<span class="_ _5"></span>,<span class="_ _6"> </span>the<span class="_ _a"> </span>strategy<span class="_ _6"> </span>in<span class="_ _a"> </span>Ciresan<span class="_ _6"> </span>et<span class="_ _6"> </span>al.<span class="_ _a"> </span>[<span class="fc2">1</span>]<span class="_ _6"> </span>has<span class="_ _a"> </span>tw<span class="_ _1"></span>o<span class="_ _a"> </span>drawbac<span class="_ _1"></span>ks.<span class="_ _6"> </span>First,<span class="_ _a"> </span>it</div><div class="t m0 x9 h3 y5c ff2 fs1 fc0 sc0 ls0 ws0">is<span class="_ _8"> </span>quite<span class="_ _8"> </span>slow<span class="_ _8"> </span>b<span class="_ _3"></span>ecause<span class="_ _8"> </span>the<span class="_ _9"> </span>net<span class="_ _1"></span>work<span class="_ _7"> </span>must<span class="_ _8"> </span>b<span class="_ _3"></span>e<span class="_ _8"> </span>run<span class="_ _9"> </span>separately<span class="_ _8"> </span>for<span class="_ _8"> </span>each<span class="_ _8"> </span>patc<span class="_ _1"></span>h,<span class="_ _8"> </span>and</div><div class="t m0 x9 h3 y5d ff2 fs1 fc0 sc0 ls0 ws0">there<span class="_ _9"> </span>is<span class="_ _a"> </span>a<span class="_ _9"> </span>lot<span class="_ _9"> </span>of<span class="_ _a"> </span>redundancy<span class="_ _9"> </span>due<span class="_ _a"> </span>to<span class="_ _9"> </span>ov<span class="_ _1"></span>erlapping<span class="_ _9"> </span>patches.<span class="_ _9"> </span>Secondly<span class="_ _5"></span>,<span class="_ _9"> </span>there<span class="_ _a"> </span>is<span class="_ _9"> </span>a</div><div class="t m0 x9 h3 y5e ff2 fs1 fc0 sc0 ls0 ws0">trade-o&#64256;<span class="_ _8"> </span>b<span class="_ _3"></span>et<span class="_ _1"></span>ween<span class="_ _7"> </span>lo<span class="_ _3"></span>calization<span class="_ _8"> </span>accuracy<span class="_ _8"> </span>and<span class="_ _8"> </span>the<span class="_ _8"> </span>use<span class="_ _8"> </span>of<span class="_ _8"> </span>context.<span class="_ _7"> </span>Larger<span class="_ _8"> </span>patches</div><div class="t m0 x9 h3 y5f ff2 fs1 fc0 sc0 ls0 ws0">require<span class="_ _a"> </span>more<span class="_ _6"> </span>max-po<span class="_ _3"></span>oling<span class="_ _6"> </span>la<span class="_ _1"></span>yers<span class="_ _a"> </span>that<span class="_ _a"> </span>reduce<span class="_ _6"> </span>the<span class="_ _a"> </span>lo<span class="_ _3"></span>calization<span class="_ _6"> </span>accuracy<span class="_ _5"></span>,<span class="_ _a"> </span>while</div><div class="t m0 x9 h3 y60 ff2 fs1 fc0 sc0 ls0 ws0">small<span class="_ _10"> </span>patches<span class="_ _10"> </span>allo<span class="_ _1"></span>w<span class="_ _c"> </span>the<span class="_ _10"> </span>netw<span class="_ _1"></span>ork<span class="_ _10"> </span>to<span class="_ _c"> </span>see<span class="_ _10"> </span>only<span class="_ _c"> </span>little<span class="_ _10"> </span>context.<span class="_ _10"> </span>More<span class="_ _10"> </span>recent<span class="_ _10"> </span>approac<span class="_ _1"></span>hes</div><div class="t m0 x9 h3 y61 ff2 fs1 fc0 sc0 ls0 ws0">[<span class="fc2">11</span>,<span class="fc2">4</span>]<span class="_ _6"> </span>proposed<span class="_ _6"> </span>a<span class="_ _6"> </span>classi&#64257;er<span class="_ _a"> </span>output<span class="_ _6"> </span>that<span class="_ _6"> </span>tak<span class="_ _1"></span>es<span class="_ _6"> </span>in<span class="_ _1"></span>to<span class="_ _6"> </span>account<span class="_ _a"> </span>the<span class="_ _6"> </span>features<span class="_ _a"> </span>from</div><div class="t m0 x9 h3 y62 ff2 fs1 fc0 sc0 ls0 ws0">m<span class="_ _1"></span>ultiple<span class="_ _a"> </span>lay<span class="_ _1"></span>ers.<span class="_ _a"> </span>Go<span class="_ _3"></span>o<span class="_ _3"></span>d<span class="_ _a"> </span>lo<span class="_ _3"></span>calization<span class="_ _a"> </span>and<span class="_ _a"> </span>the<span class="_ _a"> </span>use<span class="_ _a"> </span>of<span class="_ _a"> </span>con<span class="_ _1"></span>text<span class="_ _a"> </span>are<span class="_ _a"> </span>p<span class="_ _3"></span>ossible<span class="_ _a"> </span>at<span class="_ _a"> </span>the</div><div class="t m0 x9 h3 y63 ff2 fs1 fc0 sc0 ls0 ws0">same<span class="_ _2"> </span>time.</div><div class="t m0 xa h3 y21 ff2 fs1 fc0 sc0 ls0 ws0">In<span class="_"> </span>this<span class="_ _4"> </span>pap<span class="_ _3"></span>er,<span class="_ _4"> </span>w<span class="_ _1"></span>e<span class="_ _4"> </span>build<span class="_ _4"> </span>up<span class="_ _3"></span>on<span class="_"> </span>a<span class="_ _4"> </span>more<span class="_ _4"> </span>elegan<span class="_ _1"></span>t<span class="_ _4"> </span>architecture,<span class="_"> </span>the<span class="_"> </span>so-called<span class="_ _4"> </span>&#8220;fully</div><div class="t m0 x9 h3 y22 ff2 fs1 fc0 sc0 ls0 ws0">con<span class="_ _1"></span>volutional<span class="_"> </span>netw<span class="_ _1"></span>ork&#8221;<span class="_ _4"> </span>[<span class="fc2">9</span>].<span class="_ _4"> </span>W<span class="_ _5"></span>e<span class="_ _4"> </span>mo<span class="_ _3"></span>dify<span class="_ _4"> </span>and<span class="_ _4"> </span>extend<span class="_ _2"> </span>this<span class="_ _4"> </span>architecture<span class="_"> </span>such<span class="_"> </span>that<span class="_ _4"> </span>it</div><div class="t m0 x9 h3 y23 ff2 fs1 fc0 sc0 ls0 ws0">w<span class="_ _1"></span>orks<span class="_ _2"> </span>with<span class="_ _2"> </span>very<span class="_ _4"> </span>few<span class="_ _4"> </span>training<span class="_ _2"> </span>images<span class="_ _2"> </span>and<span class="_ _2"> </span>yields<span class="_ _4"> </span>more<span class="_ _2"> </span>precise<span class="_ _2"> </span>segmentations;<span class="_ _4"> </span>see</div><div class="t m0 x9 h3 y24 ff2 fs1 fc2 sc0 ls0 ws0">Figure<span class="_ _2"> </span>1<span class="fc0">.<span class="_ _2"> </span>The<span class="_ _7"> </span>main<span class="_ _2"> </span>idea<span class="_ _2"> </span>in<span class="_ _7"> </span>[</span>9<span class="fc0">]<span class="_ _2"> </span>is<span class="_ _2"> </span>to<span class="_ _7"> </span>supplement<span class="_ _4"> </span>a<span class="_ _2"> </span>usual<span class="_ _7"> </span>con<span class="_ _1"></span>tracting<span class="_ _7"> </span>net<span class="_ _1"></span>work<span class="_ _4"> </span>by</span></div><div class="t m0 x9 h3 y25 ff2 fs1 fc0 sc0 ls0 ws0">successiv<span class="_ _1"></span>e<span class="_"> </span>lay<span class="_ _1"></span>ers,<span class="_"> </span>where<span class="_ _c"> </span>p<span class="_ _3"></span>o<span class="_ _3"></span>oling<span class="_"> </span>operators<span class="_"> </span>are<span class="_"> </span>replaced<span class="_"> </span>b<span class="_ _1"></span>y<span class="_"> </span>upsampling<span class="_"> </span>operators.</div><div class="t m0 x9 h3 y26 ff2 fs1 fc0 sc0 ls0 ws0">Hence,<span class="_ _10"> </span>these<span class="_"> </span>la<span class="_ _1"></span>y<span class="_ _1"></span>ers<span class="_ _c"> </span>increase<span class="_ _c"> </span>the<span class="_ _10"> </span>resolution<span class="_"> </span>of<span class="_ _10"> </span>the<span class="_ _10"> </span>output.<span class="_"> </span>In<span class="_ _10"> </span>order<span class="_ _10"> </span>to<span class="_"> </span>localize,<span class="_ _c"> </span>high</div><div class="t m0 x9 h3 y27 ff2 fs1 fc0 sc0 ls0 ws0">resolution<span class="_ _4"> </span>features<span class="_ _4"> </span>from<span class="_ _2"> </span>the<span class="_ _4"> </span>contracting<span class="_ _4"> </span>path<span class="_ _4"> </span>are<span class="_ _4"> </span>combined<span class="_ _4"> </span>with<span class="_ _4"> </span>the<span class="_ _4"> </span>upsampled</div><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a></div><div class="pi" data-data='{"ctm":[1.568627,0.000000,0.000000,1.568627,0.000000,0.000000]}'></div></div><div id="pf3" class="pf w0 h0" data-page-no="3"><div class="pc pc3 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/89912414/bg3.jpg"><div class="t m0 x34 h4 y29 ff3 fs2 fc0 sc0 ls0 ws0">3</div><div class="t m0 x9 h4 y64 ff5 fs2 fc0 sc0 ls0 ws0">Fig.<span class="_ _f"> </span>2.<span class="_ _4"> </span><span class="ff3">Ov<span class="_ _1"></span>erlap-tile<span class="_ _4"> </span>strategy<span class="_"> </span>for<span class="_ _4"> </span>seamless<span class="_ _4"> </span>segmen<span class="_ _1"></span>tation<span class="_ _4"> </span>of<span class="_"> </span>arbitrary<span class="_ _4"> </span>large<span class="_ _4"> </span>images<span class="_"> </span>(here</span></div><div class="t m0 x9 h4 y65 ff3 fs2 fc0 sc0 ls0 ws0">segmen<span class="_ _1"></span>tation<span class="_ _2"> </span>of<span class="_ _4"> </span>neuronal<span class="_ _2"> </span>structures<span class="_ _4"> </span>in<span class="_ _2"> </span>EM<span class="_ _4"> </span>stacks).<span class="_ _4"> </span>Prediction<span class="_ _4"> </span>of<span class="_ _2"> </span>the<span class="_ _4"> </span>segmentation<span class="_ _4"> </span>in</div><div class="t m0 x9 h4 y66 ff3 fs2 fc0 sc0 ls0 ws0">the<span class="_ _4"> </span>y<span class="_ _1"></span>ellow<span class="_"> </span>area,<span class="_ _4"> </span>requires<span class="_ _4"> </span>image<span class="_ _4"> </span>data<span class="_"> </span>within<span class="_ _4"> </span>the<span class="_ _4"> </span>blue<span class="_ _4"> </span>area<span class="_"> </span>as<span class="_ _4"> </span>input.<span class="_ _4"> </span>Missing<span class="_ _4"> </span>input<span class="_"> </span>data</div><div class="t m0 x9 h4 y67 ff3 fs2 fc0 sc0 ls0 ws0">is<span class="_ _4"> </span>extrap<span class="_ _3"></span>olated<span class="_ _4"> </span>b<span class="_ _1"></span>y<span class="_ _4"> </span>mirroring</div><div class="t m0 x9 h3 y68 ff2 fs1 fc0 sc0 ls0 ws0">output.<span class="_"> </span>A<span class="_"> </span>successive<span class="_"> </span>con<span class="_ _1"></span>volution<span class="_"> </span>la<span class="_ _1"></span>y<span class="_ _1"></span>er<span class="_"> </span>can<span class="_"> </span>then<span class="_ _4"> </span>learn<span class="_"> </span>to<span class="_"> </span>assemble<span class="_"> </span>a<span class="_"> </span>more<span class="_"> </span>precise</div><div class="t m0 x9 h3 y69 ff2 fs1 fc0 sc0 ls0 ws0">output<span class="_ _2"> </span>based<span class="_ _2"> </span>on<span class="_ _2"> </span>this<span class="_ _2"> </span>information.</div><div class="t m0 xa h3 y6a ff2 fs1 fc0 sc0 ls0 ws0">One<span class="_ _9"> </span>imp<span class="_ _3"></span>ortant<span class="_ _9"> </span>modi&#64257;cation<span class="_ _9"> </span>in<span class="_ _a"> </span>our<span class="_ _9"> </span>architecture<span class="_ _9"> </span>is<span class="_ _9"> </span>that<span class="_ _9"> </span>in<span class="_ _9"> </span>the<span class="_ _9"> </span>upsampling</div><div class="t m0 x9 h3 y6b ff2 fs1 fc0 sc0 ls0 ws0">part<span class="_ _7"> </span>we<span class="_ _2"> </span>hav<span class="_ _1"></span>e<span class="_ _7"> </span>also<span class="_ _7"> </span>a<span class="_ _7"> </span>large<span class="_ _7"> </span>num<span class="_ _1"></span>b<span class="_ _3"></span>er<span class="_ _7"> </span>of<span class="_ _7"> </span>feature<span class="_ _7"> </span>channels,<span class="_ _2"> </span>which<span class="_ _2"> </span>allow<span class="_ _2"> </span>the<span class="_ _8"> </span>net<span class="_ _1"></span>work</div><div class="t m0 x9 h3 y6c ff2 fs1 fc0 sc0 ls0 ws0">to<span class="_ _4"> </span>propagate<span class="_ _2"> </span>context<span class="_ _4"> </span>information<span class="_ _2"> </span>to<span class="_ _2"> </span>higher<span class="_ _4"> </span>resolution<span class="_ _2"> </span>lay<span class="_ _1"></span>ers.<span class="_ _2"> </span>As<span class="_ _4"> </span>a<span class="_ _2"> </span>consequence,</div><div class="t m0 x9 h3 y6d ff2 fs1 fc0 sc0 ls0 ws0">the<span class="_"> </span>expansive<span class="_"> </span>path<span class="_"> </span>is<span class="_"> </span>more<span class="_ _4"> </span>or<span class="_"> </span>less<span class="_ _4"> </span>symmetric<span class="_"> </span>to<span class="_ _4"> </span>the<span class="_"> </span>contracting<span class="_"> </span>path,<span class="_"> </span>and<span class="_"> </span>yields</div><div class="t m0 x9 h3 y6e ff2 fs1 fc0 sc0 ls0 ws0">a<span class="_ _7"> </span>u-shap<span class="_ _3"></span>ed<span class="_ _7"> </span>architecture.<span class="_ _2"> </span>The<span class="_ _7"> </span>netw<span class="_ _1"></span>ork<span class="_ _2"> </span>do<span class="_ _3"></span>es<span class="_ _7"> </span>not<span class="_ _8"> </span>ha<span class="_ _1"></span>ve<span class="_ _2"> </span>any<span class="_ _2"> </span>fully<span class="_ _7"> </span>connected<span class="_ _7"> </span>lay<span class="_ _1"></span>ers</div><div class="t m0 x9 h3 y6f ff2 fs1 fc0 sc0 ls0 ws0">and<span class="_"> </span>only<span class="_"> </span>uses<span class="_"> </span>the<span class="_"> </span>v<span class="_ _1"></span>alid<span class="_"> </span>part<span class="_"> </span>of<span class="_"> </span>each<span class="_"> </span>con<span class="_ _1"></span>volution,<span class="_"> </span>i.e.,<span class="_"> </span>the<span class="_"> </span>segmen<span class="_ _1"></span>tation<span class="_"> </span>map<span class="_"> </span>only</div><div class="t m0 x9 h3 y70 ff2 fs1 fc0 sc0 ls0 ws0">con<span class="_ _1"></span>tains<span class="_ _9"> </span>the<span class="_ _8"> </span>pixels,<span class="_ _9"> </span>for<span class="_ _9"> </span>whic<span class="_ _1"></span>h<span class="_ _9"> </span>the<span class="_ _8"> </span>full<span class="_ _9"> </span>con<span class="_ _1"></span>text<span class="_ _9"> </span>is<span class="_ _8"> </span>av<span class="_ _5"></span>ailable<span class="_ _9"> </span>in<span class="_ _8"> </span>the<span class="_ _9"> </span>input<span class="_ _8"> </span>image.</div><div class="t m0 x9 h3 y71 ff2 fs1 fc0 sc0 ls0 ws0">This<span class="_"> </span>strategy<span class="_ _4"> </span>allows<span class="_"> </span>the<span class="_ _4"> </span>seamless<span class="_"> </span>segmentation<span class="_"> </span>of<span class="_ _4"> </span>arbitrarily<span class="_ _4"> </span>large<span class="_"> </span>images<span class="_ _4"> </span>by<span class="_"> </span>an</div><div class="t m0 x9 h3 y72 ff2 fs1 fc0 sc0 ls0 ws0">o<span class="_ _1"></span>verlap-tile<span class="_ _8"> </span>strategy<span class="_ _8"> </span>(see<span class="_ _9"> </span><span class="fc2">Figure<span class="_ _8"> </span>2</span>).<span class="_ _8"> </span>T<span class="_ _5"></span>o<span class="_ _8"> </span>predict<span class="_ _9"> </span>the<span class="_ _8"> </span>pixels<span class="_ _8"> </span>in<span class="_ _8"> </span>the<span class="_ _9"> </span>border<span class="_ _9"> </span>region</div><div class="t m0 x9 h3 y73 ff2 fs1 fc0 sc0 ls0 ws0">of<span class="_ _4"> </span>the<span class="_ _2"> </span>image,<span class="_ _2"> </span>the<span class="_ _2"> </span>missing<span class="_ _4"> </span>context<span class="_ _4"> </span>is<span class="_ _2"> </span>extrap<span class="_ _3"></span>olated<span class="_ _2"> </span>by<span class="_ _4"> </span>mirroring<span class="_ _4"> </span>the<span class="_ _2"> </span>input<span class="_ _2"> </span>image.</div><div class="t m0 x9 h3 y74 ff2 fs1 fc0 sc0 ls0 ws0">This<span class="_ _9"> </span>tiling<span class="_ _a"> </span>strategy<span class="_ _9"> </span>is<span class="_ _a"> </span>imp<span class="_ _3"></span>ortan<span class="_ _1"></span>t<span class="_ _a"> </span>to<span class="_ _9"> </span>apply<span class="_ _a"> </span>the<span class="_ _9"> </span>netw<span class="_ _1"></span>ork<span class="_ _9"> </span>to<span class="_ _a"> </span>large<span class="_ _9"> </span>images,<span class="_ _a"> </span>since</div><div class="t m0 x9 h3 y75 ff2 fs1 fc0 sc0 ls0 ws0">otherwise<span class="_ _2"> </span>the<span class="_ _2"> </span>resolution<span class="_ _2"> </span>w<span class="_ _1"></span>ould<span class="_ _2"> </span>b<span class="_ _3"></span>e<span class="_ _2"> </span>limited<span class="_ _2"> </span>by<span class="_ _4"> </span>the<span class="_ _2"> </span>GPU<span class="_ _2"> </span>memory<span class="_ _5"></span>.</div><div class="t m0 xa h3 y76 ff2 fs1 fc0 sc0 ls0 ws0">As<span class="_ _2"> </span>for<span class="_ _7"> </span>our<span class="_ _7"> </span>tasks<span class="_ _2"> </span>there<span class="_ _7"> </span>is<span class="_ _7"> </span>v<span class="_ _1"></span>ery<span class="_ _7"> </span>little<span class="_ _7"> </span>training<span class="_ _2"> </span>data<span class="_ _7"> </span>av<span class="_ _5"></span>ailable,<span class="_ _7"> </span>w<span class="_ _1"></span>e<span class="_ _7"> </span>use<span class="_ _7"> </span>excessiv<span class="_ _1"></span>e</div><div class="t m0 x9 h3 y77 ff2 fs1 fc0 sc0 ls0 ws0">data<span class="_"> </span>augmen<span class="_ _1"></span>tation<span class="_"> </span>b<span class="_ _1"></span>y<span class="_"> </span>applying<span class="_"> </span>elastic<span class="_"> </span>deformations<span class="_ _c"> </span>to<span class="_"> </span>the<span class="_"> </span>a<span class="_ _1"></span>v<span class="_ _5"></span>ailable<span class="_"> </span>training<span class="_"> </span>im-</div><div class="t m0 x9 h3 y78 ff2 fs1 fc0 sc0 ls0 ws0">ages.<span class="_ _4"> </span>This<span class="_ _2"> </span>allows<span class="_ _4"> </span>the<span class="_ _4"> </span>netw<span class="_ _1"></span>ork<span class="_ _4"> </span>to<span class="_ _2"> </span>learn<span class="_ _4"> </span>inv<span class="_ _5"></span>ariance<span class="_ _2"> </span>to<span class="_ _2"> </span>such<span class="_"> </span>deformations,<span class="_ _2"> </span>without</div><div class="t m0 x9 h3 y79 ff2 fs1 fc0 sc0 ls0 ws0">the<span class="_ _9"> </span>need<span class="_ _9"> </span>to<span class="_ _9"> </span>see<span class="_ _8"> </span>these<span class="_ _9"> </span>transformations<span class="_ _9"> </span>in<span class="_ _9"> </span>the<span class="_ _9"> </span>annotated<span class="_ _9"> </span>image<span class="_ _9"> </span>corpus.<span class="_ _9"> </span>This<span class="_ _9"> </span>is</div><div class="t m0 x9 h3 y7a ff2 fs1 fc0 sc0 ls0 ws0">particularly<span class="_ _9"> </span>imp<span class="_ _3"></span>ortant<span class="_ _8"> </span>in<span class="_ _9"> </span>biomedical<span class="_ _a"> </span>segmen<span class="_ _1"></span>tation,<span class="_ _9"> </span>since<span class="_ _9"> </span>deformation<span class="_ _9"> </span>used<span class="_ _a"> </span>to</div><div class="t m0 x9 h3 y7b ff2 fs1 fc0 sc0 ls0 ws0">b<span class="_ _3"></span>e<span class="_ _4"> </span>the<span class="_ _4"> </span>most<span class="_ _2"> </span>common<span class="_ _4"> </span>v<span class="_ _1"></span>ariation<span class="_ _4"> </span>in<span class="_ _4"> </span>tissue<span class="_ _2"> </span>and<span class="_ _4"> </span>realistic<span class="_ _4"> </span>deformations<span class="_ _2"> </span>can<span class="_ _4"> </span>b<span class="_ _3"></span>e<span class="_ _2"> </span>simu-</div><div class="t m0 x9 h3 y7c ff2 fs1 fc0 sc0 ls0 ws0">lated<span class="_"> </span>e&#64259;cien<span class="_ _1"></span>tly<span class="_ _5"></span>.<span class="_"> </span>The<span class="_"> </span>v<span class="_ _5"></span>alue<span class="_"> </span>of<span class="_"> </span>data<span class="_ _c"> </span>augmentation<span class="_ _10"> </span>for<span class="_"> </span>learning<span class="_"> </span>in<span class="_ _1"></span>v<span class="_ _1"></span>ariance<span class="_"> </span>has<span class="_"> </span>been</div><div class="t m0 x9 h3 y7d ff2 fs1 fc0 sc0 ls0 ws0">sho<span class="_ _1"></span>wn<span class="_ _2"> </span>in<span class="_ _2"> </span>Dosovitskiy<span class="_ _4"> </span>et<span class="_ _2"> </span>al.<span class="_ _2"> </span>[<span class="fc2">2</span>]<span class="_ _2"> </span>in<span class="_ _2"> </span>the<span class="_ _2"> </span>scop<span class="_ _3"></span>e<span class="_ _2"> </span>of<span class="_ _2"> </span>unsup<span class="_ _3"></span>ervised<span class="_ _2"> </span>feature<span class="_ _2"> </span>learning.</div><div class="t m0 xa h3 y7e ff2 fs1 fc0 sc0 ls0 ws0">Another<span class="_"> </span>c<span class="_ _1"></span>hallenge<span class="_"> </span>in<span class="_ _c"> </span>many<span class="_ _c"> </span>cell<span class="_"> </span>segmen<span class="_ _1"></span>tation<span class="_"> </span>tasks<span class="_"> </span>is<span class="_ _10"> </span>the<span class="_"> </span>separation<span class="_"> </span>of<span class="_"> </span>touc<span class="_ _1"></span>h-</div><div class="t m0 x9 h3 y7f ff2 fs1 fc0 sc0 ls0 ws0">ing<span class="_ _7"> </span>ob<span class="_ _11"></span>jects<span class="_ _7"> </span>of<span class="_ _8"> </span>the<span class="_ _8"> </span>same<span class="_ _7"> </span>class;<span class="_ _8"> </span>see<span class="_ _8"> </span><span class="fc2">Figure<span class="_ _7"> </span>3</span>.<span class="_ _8"> </span>T<span class="_ _5"></span>o<span class="_ _8"> </span>this<span class="_ _7"> </span>end,<span class="_ _8"> </span>we<span class="_ _2"> </span>prop<span class="_ _3"></span>ose<span class="_ _8"> </span>the<span class="_ _8"> </span>use<span class="_ _7"> </span>of</div><div class="t m0 x9 h3 y80 ff2 fs1 fc0 sc0 ls0 ws0">a<span class="_ _2"> </span>weigh<span class="_ _1"></span>ted<span class="_ _2"> </span>loss,<span class="_ _2"> </span>where<span class="_ _7"> </span>the<span class="_ _2"> </span>separating<span class="_ _2"> </span>background<span class="_ _2"> </span>lab<span class="_ _3"></span>els<span class="_ _2"> </span>b<span class="_ _3"></span>etw<span class="_ _1"></span>een<span class="_ _2"> </span>touching<span class="_ _4"> </span>cells</div><div class="t m0 x9 h3 y81 ff2 fs1 fc0 sc0 ls0 ws0">obtain<span class="_ _2"> </span>a<span class="_ _2"> </span>large<span class="_ _2"> </span>w<span class="_ _1"></span>eight<span class="_ _4"> </span>in<span class="_ _2"> </span>the<span class="_ _2"> </span>loss<span class="_ _2"> </span>function.</div><div class="t m0 xa h3 y25 ff2 fs1 fc0 sc0 ls0 ws0">The<span class="_"> </span>resulting<span class="_"> </span>net<span class="_ _1"></span>w<span class="_ _1"></span>ork<span class="_"> </span>is<span class="_"> </span>applicable<span class="_"> </span>to<span class="_"> </span>v<span class="_ _5"></span>arious<span class="_"> </span>biomedical<span class="_"> </span>segmen<span class="_ _1"></span>tation<span class="_"> </span>prob-</div><div class="t m0 x9 h3 y26 ff2 fs1 fc0 sc0 ls0 ws0">lems.<span class="_ _4"> </span>In<span class="_ _2"> </span>this<span class="_ _2"> </span>pap<span class="_ _3"></span>er,<span class="_ _2"> </span>we<span class="_ _4"> </span>sho<span class="_ _1"></span>w<span class="_ _2"> </span>results<span class="_ _2"> </span>on<span class="_ _4"> </span>the<span class="_ _2"> </span>segmentation<span class="_ _4"> </span>of<span class="_ _2"> </span>neuronal<span class="_ _4"> </span>structures</div><div class="t m0 x9 h3 y27 ff2 fs1 fc0 sc0 ls0 ws0">in<span class="_ _a"> </span>EM<span class="_ _a"> </span>stac<span class="_ _1"></span>ks<span class="_ _a"> </span>(an<span class="_ _a"> </span>ongoing<span class="_ _a"> </span>comp<span class="_ _3"></span>etition<span class="_ _a"> </span>started<span class="_ _a"> </span>at<span class="_ _a"> </span>ISBI<span class="_ _a"> </span>2012),<span class="_ _a"> </span>where<span class="_ _a"> </span>we<span class="_ _9"> </span>out-</div><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a><a class="l"><div class="d m2"></div></a></div><div class="pi" data-data='{"ctm":[1.568627,0.000000,0.000000,1.568627,0.000000,0.000000]}'></div></div>
100+评论
captcha
    相关资源
    类型标题大小时间
    ZIP基于Java的视频会议系统（程序）.zip3.63MB6月前
    ZIP功能特点：支持上传较大体积的 Word 文档（最大支持上传 48mb），压缩比较高 该网站无广告，使用方便，并且支持痕迹清除，注5.1MB6月前
    ZIPVMware 脚本存储库，该存储库包含基于 VMware 的解决方案的各种编程/脚本语言的示例脚本集合815.55KB6月前
    ZIPJAVA在线考试管理系统(源代码).zip1.19MB6月前
    ZIPKafka Tools 3.0，现改名 offsetexplorer86.22MB6月前
    ZIPpython pywin32阅读文档 python的pywin32api帮助文档2.29MB6月前
    ZIPredis-desktop-manager-Redis桌面管理器26.84MB6月前
    ZIPGAN网络的几篇论文，主要是图像方面的论文39.68MB6月前