{
  "version": "https://jsonfeed.org/version/1", 
  "title": "Caffe", 
  "description": "Caffe is a deep learning framework made with expression, speed, and modularity in mind.", 
  "home_page_url": "https://www.v2ex.com/go/caffe", 
  "feed_url": "https://www.v2ex.com/feed/caffe.json", 
  "icon": "https://cdn.v2ex.com/navatar/9348/15ad/994_large.png?m=1496947126", 
  "favicon": "https://cdn.v2ex.com/navatar/9348/15ad/994_normal.png?m=1496947126", 
  "items": [
    {
      "author": {
        "url": "https://www.v2ex.com/member/benzzz", 
        "name": "benzzz", 
        "avatar": "https://cdn.v2ex.com/avatar/70c3/2807/279697_large.png?m=1542157612"
      }, 
      "url": "https://www.v2ex.com/t/632249", 
      "title": "\u79cd\u8349\u4e86 GENE CAFE \u70d8\u8c46\u673a\uff0c\u60f3\u5c1d\u8bd5\u4e00\u4e0b\u81ea\u5df1\u70d8\u8c46\uff0c\u5404\u4f4d V \u53cb\u6709\u4f7f\u7528\u8fc7\u8fd9\u53f0\u673a\u5668\u7684\u5417\uff1f", 
      "id": "https://www.v2ex.com/t/632249", 
      "date_published": "2019-12-25T09:13:12+00:00", 
      "content_html": "<p>\u4ece\u53bb\u5e74\u5165\u5751\u624b\u51b2\u5496\u5561\u5f00\u59cb\uff0c\u4e00\u6b65\u6b65\u5347\u7ea7\u5668\u5177\uff0c\u6162\u6162\u5730\u6709\u4e86\u5165\u5751\u81ea\u5df1\u70d8\u8c46\u5b50\u60f3\u6cd5\uff0c\u5c24\u5176\u662f\u6628\u665a\u770b\u4e86\u624b\u7f51\u70d8\u7119\u5496\u5561\u8c46\u7684\u51e0\u4e2a\u89c6\u9891\uff0c\u66f4\u662f\u6709\u4e86\u5f3a\u70c8\u7684\u5165\u5751\u60f3\u6cd5\u3002</p>\n<p>\u8bb2\u9053\u7406\u5148\u4ece\u624b\u7f51\u5f00\u59cb\u5165\u5751\u662f\u4e00\u4e2a\u5f88\u4e0d\u9519\u7684\u9009\u62e9\uff0c\u8bd5\u9519\u6210\u672c\u4f4e\uff0c\u80fd\u66f4\u76f4\u89c2\u5730\u4f53\u9a8c\u5230\u70d8\u7119\u7684\u4e50\u8da3\u3002\u4f46\u662f\u7f3a\u70b9\u4e5f\u5f88\u660e\u663e\uff1a\u5473\u9053\u4f1a\u6bd4\u8f83\u5927\uff1b\u70d8\u7119\u8fc7\u7a0b\u4e2d\u5230\u5904\u98d8\u98de\u5730\u94f6\u76ae\uff0c\u5728\u540e\u9762\u6e05\u7406\u7684\u65f6\u5019\u4f1a\u6bd4\u8f83\u5934\u75bc\uff08\u8fd9\u4e2a\u771f\u7684\u4f1a\u88ab\u5bb6\u91cc\u4eba\u7ed9\u5acc\u5f03\uff09\uff1b\u5bf9\u81c2\u529b\u8981\u6c42\u6bd4\u8f83\u9ad8\uff1b\u5355\u6b21\u51fa\u54c1\u5206\u91cf\u4e0d\u591a\u3002</p>\n<p>\u540e\u6765\u5728\u4e0d\u65ad\u5730\u641c\u7d22\u4e0b\uff0c\u627e\u5230\u4e86 GENE CAFE \u8fd9\u6b3e\u5c0f\u578b\u70d8\u8c46\u673a\uff0c\u4ef7\u683c\u8fd8\u884c\uff0c\u67d0\u5b9d 3k+\uff0c\u641c\u7d22\u8fd9\u6b3e\u673a\u5668\u56fd\u5185\u4ea4\u6d41\u5fc3\u5f97\u53d1\u73b0\u5e16\u5b50\u6bd4\u8f83\u5c11\uff0c\u6709\u4eba\u6307\u51fa\u8fd9\u6b3e\u673a\u5668\u8fd8\u662f\u4e00\u4e2a\u70d8\u7119\u6548\u679c\u4e00\u822c\u7684\u73a9\u5177\uff0c\u5728\u4e8e\u5347\u6e29\u8f83\u6162\uff0c\u6700\u540e\u51fa\u54c1\u4e0d\u4f1a\u5f88\u9ad8\u3002</p>\n<p>\u4e0d\u77e5\u9053 V \u53cb\u6709\u6ca1\u6709\u81ea\u884c\u70d8\u7119\u8c46\u5b50\u7684\u7231\u597d\uff0c\u53ef\u4ee5\u4ea4\u6d41\u4ea4\u6d41\uff0c\u8ba9\u6211\u5b66\u4e60\u4e00\u4e0b\u3002</p>\n"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/fullStackDude", 
        "name": "fullStackDude", 
        "avatar": "https://cdn.v2ex.com/avatar/cd37/ac7f/181979_large.png?m=1472540539"
      }, 
      "url": "https://www.v2ex.com/t/361181", 
      "date_modified": "2017-06-09T13:29:08+00:00", 
      "content_html": "<p>\u4e1a\u4f59\u65f6\u95f4\u5b9e\u73b0\u7684\u4e00\u4e2a demo project\uff1a</p>\n<p><a href=\"https://github.com/KleinYuan/Caffe2-iOS\" rel=\"nofollow\">https://github.com/KleinYuan/Caffe2-iOS</a></p>\n<p>\u6b22\u8fce\u5927\u5bb6\u4e00\u8d77 contribute</p>\n", 
      "date_published": "2017-05-13T20:41:11+00:00", 
      "title": "\u5206\u4eab\u4e00\u4e2a Caffe2 on iOS \u7684 demo project", 
      "id": "https://www.v2ex.com/t/361181"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/Reign", 
        "name": "Reign", 
        "avatar": "https://cdn.v2ex.com/avatar/c277/2fe6/160960_large.png?m=1456660452"
      }, 
      "url": "https://www.v2ex.com/t/357747", 
      "date_modified": "2017-06-08T18:39:12+00:00", 
      "content_html": "caffe \u4e2d\u793a\u4f8b\u7684\u5982\u4e0b\u4ee3\u7801\uff1a\r<br />import numpy as np\r<br />import sys,os\r<br />caffe_root = '/opt/caffe/' \r<br />sys.path.insert(0, caffe_root + 'python')\r<br />import caffe\r<br />os.chdir(caffe_root)\r<br />net_file=caffe_root + 'deploy.prototxt'\r<br />caffe_model=caffe_root + 'test.caffemodel'\r<br />mean_file=caffe_root + 'mean.npy'\r<br />net = caffe.Net(net_file,caffe_model,caffe.TEST)\r<br />transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})\r<br />transformer.set_transpose('data', (2,0,1))\r<br />transformer.set_mean('data', np.load(mean_file).mean(1).mean(1))\r<br />transformer.set_raw_scale('data', 255) \r<br />transformer.set_channel_swap('data', (2,1,0))\r<br />#\u628a transformer \u6301\u4e45\u5316\u5b58\u50a8\u4e0b\u6765\r<br />im=caffe.io.load_image(caffe_root+'/sample/1.jpg')\r<br />net.blobs['data'].data[...] = transformer.preprocess('data',im)\r<br />out = net.forward()\r<br />print out\r<br />\u670d\u52a1\u5668\u6ca1 GPU\uff0c\u6bcf\u6b21\u52a0\u8f7d\u4e00\u4e2a transformer \u975e\u5e38\u8017\u65f6\uff0c\u6211\u5c31\u5728\u60f3\u53ea\u52a0\u8f7d\u4e00\u6b21\uff0c\u80fd\u4e0d\u80fd\u6301\u4e45\u5316\u628a\u8fd9\u4e2a transformer \u5b58\u50a8\u4e0b\u6765\uff0c\u4ee5\u540e\u6bcf\u6b21\u6d4b\u8bd5\u56fe\u7247\u76f4\u63a5\u8c03\u7528\u5c31\u884c\u4e86\uff1f\r<br />\u53e6\u5916\uff0c\u5982\u679c\u4f7f\u7528\u522b\u4eba\u5df2\u7ecf\u5efa\u7acb\u597d\u4e86\u7684 caffemodel \u6765\u6d4b\u8bd5\uff0c\u662f\u4e0d\u662f\u6709\u4e86 GPU \u4f1a\u66f4\u5feb\u4e9b\uff1f\r<br />\u673a\u5668\u5b66\u4e60\u65b0\u624b\u95ee\u9898\u4f30\u8ba1\u6bd4\u8f83\u5e7c\u7a1a\uff0c\u8fd8\u662f\u5148\u8c22\u8c22\u4e86", 
      "date_published": "2017-04-27T08:47:43+00:00", 
      "title": "caffe \u80fd\u4e0d\u80fd\u628a transformer \u8fdb\u884c\u201c\u6301\u4e45\u5316\u201d\u5b58\u50a8\u4e0b\u6765\uff1f\u4ee5\u514d\u6bcf\u6b21\u52a0\u8f7d\u90fd\u5f88\u8017\u65f6", 
      "id": "https://www.v2ex.com/t/357747"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/Reign", 
        "name": "Reign", 
        "avatar": "https://cdn.v2ex.com/avatar/c277/2fe6/160960_large.png?m=1456660452"
      }, 
      "url": "https://www.v2ex.com/t/356831", 
      "date_modified": "2017-06-08T18:39:40+00:00", 
      "content_html": "<p>\u5b8c\u5168\u7684 caffe \u65b0\u624b\uff0c\u7814\u7a76\u4e86\u4e00\u5929\u7684 caffe \uff0c\u867d\u7136\u7cca\u91cc\u7cca\u6d82\u7684\uff0c GitHub \u4e0a\u627e\u4e86\u4e2a\u9879\u76ee\uff1a <a href=\"https://github.com/BestiVictory/ILGnet\" rel=\"nofollow\">https://github.com/BestiVictory/ILGnet</a> \u5728\u7f51\u4e0a\u627e\u4e86\u4e2a Python \u4ee3\u7801\u7167\u732b\u753b\u864e\u5199\u6210\u5982\u4e0b\uff1a</p>\n<p>import numpy as np</p>\n<p>import matplotlib.pyplot as plt</p>\n<p>caffe_root = '/opt/caffe/'</p>\n<p>import sys</p>\n<p>sys.path.insert(0, caffe_root + 'python')</p>\n<p>import caffe</p>\n<p>MODEL_FILE = caffe_root + 'ILGnet/deploy.prototxt'</p>\n<p>PRETRAINED = caffe_root + 'ILGnet/ILGnet-AVA2.caffemodel'</p>\n<p>IMAGE_FILE = caffe_root+'examples/images/cat.jpg'</p>\n<p>mean_file=caffe_root + 'ILGnet/AVA2_mean.npy'</p>\n<p>caffe.set_mode_cpu()</p>\n<p>net = caffe.Classifier(MODEL_FILE, PRETRAINED,</p>\n<p>mean=np.load(mean_file).mean(1).mean(1),</p>\n<p>channel_swap=(2,1,0),</p>\n<p>raw_scale=255,</p>\n<p>image_dims=(227, 227))</p>\n<p>input_image = caffe.io.load_image(IMAGE_FILE)</p>\n<p>plt.imshow(input_image)</p>\n<p>prediction = net.predict([input_image])</p>\n<p>plt.plot(prediction[0])</p>\n<p>plt.show()</p>\n<p>print 'predicted class:', prediction[0].argmax()</p>\n<p>\u7136\u540e\u5c31\u51fa\u73b0\uff1a\nF0423 12:33:29.282009   172 insert_splits.cpp:35] Unknown bottom blob 'label' (layer 'loss1/loss', bottom index 1)</p>\n<p>\u5b8c\u5168\u65b0\u624b\uff0c\u4f30\u8ba1\u4ee3\u7801\u9519\u7684\u5f88\u79bb\u8c31\uff0c\u5927\u5bb6\u5c31\u522b\u5632\u7b11\u4e86\u54c8\uff0c\u60f3\u95ee\u4e00\u4e0b\u8fd9\u4e2a\u5982\u679c Python \u5b9e\u73b0\u7684\u8bdd\u8be5\u600e\u6837\u5199\uff1f V \u7ad9\u9ad8\u624b\u591a\uff0c\u5e0c\u671b\u80fd\u5f97\u5230\u89e3\u7b54\uff0c\u771f\u8bda\u8c22\u8c22~</p>\n", 
      "date_published": "2017-04-23T13:08:46+00:00", 
      "title": "\u65b0\u624b\u8bd5\u8fd0\u884c\u4e86\u4e00\u4e2a caffe \u7684 Python \u4ee3\u7801\uff0c\u51fa\u73b0\u8fd9\u4e2a\u9519\u8bef\u662f\u600e\u4e48\u56de\u4e8b\uff1f", 
      "id": "https://www.v2ex.com/t/356831"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/codeman", 
        "name": "codeman", 
        "avatar": "https://cdn.v2ex.com/avatar/6d73/d8d1/161519_large.png?m=1458904878"
      }, 
      "url": "https://www.v2ex.com/t/268873", 
      "date_modified": "2017-06-08T18:39:32+00:00", 
      "content_html": "<p>\u6700\u8fd1\u5b66 caffe \u7684 python \u63a5\u53e3\uff0c\u7f51\u4e0a\u7684\u8d44\u6599\u597d\u5c11\uff0c\u82f1\u6587\u7684\u90fd\u6ca1\u591a\u5c11\uff0c\u597d\u5fc3\u585e\u3002\u3002</p>\n", 
      "date_published": "2016-04-06T02:07:06+00:00", 
      "title": "\u8bf7\u95ee\u8c01\u6709 pycaffe \u7684\u6559\u7a0b\u6216\u8005 reference?", 
      "id": "https://www.v2ex.com/t/268873"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/xjx0524", 
        "name": "xjx0524", 
        "avatar": "https://cdn.v2ex.com/avatar/0bd2/703b/65125_large.png?m=1774495912"
      }, 
      "url": "https://www.v2ex.com/t/239637", 
      "date_modified": "2017-06-08T18:40:04+00:00", 
      "content_html": "<p>\u5bfc\u5e08\u8ba9\u7ffb\u8bd1\u90a3\u7bc7\u8bba\u6587<br>\n\u9898\u76ee Caffe: Convolutional Architecture for Fast Feature Embedding<br>\n\u5730\u5740\u5728\u8fd9 <a target=\"_blank\" rel=\"nofollow\" href=\"http://arxiv.org/abs/1408.5093\">http://arxiv.org/abs/1408.5093</a><br>\n\u770b\u82f1\u6587\u5927\u6982\u80fd\u61c2\uff0c\u4f46\u662f\u60f3\u8981\u7ffb\u8bd1\u6210\u4e2d\u6587\u5c31\u4e0d\u77e5\u9053\u600e\u4e48\u8bf4\u4e86\u3002\u3002\u3002</p>\n", 
      "date_published": "2015-11-28T08:18:35+00:00", 
      "title": "\u6709\u4eba\u7ffb\u8bd1\u8fc7 caffe \u6846\u67b6\u7684\u90a3\u7bc7\u8bba\u6587\u4e48", 
      "id": "https://www.v2ex.com/t/239637"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/skyduy", 
        "name": "skyduy", 
        "avatar": "https://cdn.v2ex.com/avatar/e648/60da/21516_large.png?m=1421555729"
      }, 
      "url": "https://www.v2ex.com/t/219721", 
      "date_modified": "2017-06-08T18:39:06+00:00", 
      "content_html": "\u5404\u4f4d\u5927\u795e\u4f60\u4eec\u597d\uff0c\u5bf9\u4e8e\u4e0b\u9762\u7684\u4e00\u4e2a layer \u6a21\u578b\uff1a\r<br />layer {\r<br />\u3000 name: &quot;conv1&quot;                  \r<br /> \u3000 type: &quot;Convolution&quot;        \r<br />  \u3000 bottom: &quot;data&quot;              \r<br />  \u3000 top: &quot;conv1&quot;                 \r<br />\r<br />  \u3000 param { lr_mult: 1 \u3000 decay_mult: 1 }\r<br />  \u3000 param { lr_mult: 2 \u3000 decay_mult: 0 }\r<br />\r<br /> \u3000 convolution_param {\r<br />   \u3000\u3000 num_output: 96              \r<br />  \u3000\u3000  kernel_size: 11              \r<br />  \u3000\u3000  stride: 4                    \r<br />\u3000\u3000   weight_filler {type: &quot;gaussian&quot;         \u3000 std: 0.01              }\r<br />   \u3000\u3000 bias_filler \u3000 {type: &quot;constant&quot;      \u3000 value: 0}\r<br />  \u3000}\r<br />}\r<br />\r<br />\u4e4b\u524d\u6211\u770b\u4e86 UFLDL \u6559\u7a0b\uff0c\u90a3\u91cc filter \u662f\u4e8b\u5148\u901a\u8fc7\u540c\u7c7b\u7684 training set \u7ecf\u8fc7 Sparse Autoencoder \u7684\u5f97\u51fa\u7684\uff08\u6a21\u62df\u4eba\u7c7b\u8bc6\u522b\u56fe\u50cf\u65f6\uff0c\u53ea\u6709\u5c11\u90e8\u5206\u795e\u7ecf\u6d3b\u8dc3\uff09\uff0c\u81ea\u7f16\u7801\u65f6 loss function \u4e2d\u8fd8\u6709\u4e2a KL_divergence \u5224\u7f5a\uff0c\u4f46\u8fd9\u91cc\u4f7f\u7528\u7684 loss function \u6ca1\u6709\u8be5\u9879\u3002\r<br />\u6b64\u5916\uff0c\u5bf9 filter \u8fdb\u884c\u914d\u7f6e\u7684 weight_filler \u548c bias_filler \u662f\u4ec0\u4e48\u4f5c\u7528\uff1f\u662f\u8fdb\u884c\u521d\u59cb\u5316\u7684\u5417\uff1f\r<br />\u3000\u5982\u679c\u662f\uff0c\u90a3 weight_filter \u91cc\u9762\u7684 type \u53c8\u662f\u5e72\u561b\u7684\uff1f\r<br />\u3000\u5982\u679c\u4e0d\u662f\uff0c loss function \u548c solvers \u90fd\u5df2\u7ecf\u7ed9\u51fa\u4e86\uff0c\u8fd9\u4e00\u9879\u53c8\u6709\u4ec0\u4e48\u4f5c\u7528\uff1f", 
      "date_published": "2015-09-10T08:50:04+00:00", 
      "title": "\u5173\u4e8e caffe \u4e2d\u5377\u79ef\u5c42 filter \u7684\u4e00\u4e2a\u5c0f\u7591\u95ee\uff0c\u6c42\u89e3\u7b54\uff5e\uff5e\uff5e\uff5e", 
      "id": "https://www.v2ex.com/t/219721"
    }, 
    {
      "author": {
        "url": "https://www.v2ex.com/member/askfermi", 
        "name": "askfermi", 
        "avatar": "https://cdn.v2ex.com/avatar/41f1/a4f7/68592_large.png?m=1755400741"
      }, 
      "url": "https://www.v2ex.com/t/198797", 
      "date_modified": "2017-06-08T18:39:55+00:00", 
      "content_html": "<p>\u5728\u65b0\u7248\u672c\u7684Caffe\u91cc\u7f16\u8bd1rcnn\uff0c\u63d0\u793aUnknown Command :set_phase_test<br>\n\u4f46\u662f\u8fd9\u660e\u660e\u53ea\u662f\u4e00\u4e2ademo\u7a0b\u5e8f\u554a...</p>\n\n<p><img src=\"https://cloud.githubusercontent.com/assets/10499297/8153948/53bd59ca-1367-11e5-838f-a3b88fe521d3.png\" alt=\"image\"></p>\n\n<p>Google\u4e86\u4e00\u4e0b\u90fd\u6ca1\u6709\u627e\u5230\u76f8\u5173\u7684\u5185\u5bb9\u2026\u2026</p>\n\n<p>\u5982\u679c\u6ce8\u91ca\u6389\u8fd9\u53e5\uff0c\u7a0b\u5e8f\u4f3c\u4e4e\u662f\u53ef\u4ee5\u8fd0\u884c\u7684\u3002\u4f46\u662f\u5f88\u5feb\uff0c\u5728Extracting\u4e4b\u540eMatlab\u5c31\u4f1a\u5d29\u6e83\u3002</p>\n", 
      "date_published": "2015-06-15T13:53:57+00:00", 
      "title": "Caffe \u91cc set_phase_test \u662f\u5728\u65b0\u7248\u672c\u66f4\u540d\u4e86\u5417\uff1f", 
      "id": "https://www.v2ex.com/t/198797"
    }
  ]
}