求助！！！！！！caffe的错误 Data layer prefetch queue empty

Reply to 求助！！！！！！caffe的错误 Data layer prefetch queue empty on Mon, 18 Feb 2019 04:55:44 GMT

刘看山 — Mon, 18 Feb 2019 04:55:44 GMT

数据无法读入，错误显示：

I0218 12:09:09.077704 27426 upgrade_proto.cpp:79] Attempting to upgrade batch norm layers using deprecated params: pretrained/mobilenet_iter_73000.caffemodel
I0218 12:09:09.077736 27426 upgrade_proto.cpp:82] Successfully upgraded batch norm layers using deprecated params.
I0218 12:09:09.081023 27426 net.cpp:744] Ignoring source layer conv11_mbox_conf
I0218 12:09:09.081049 27426 net.cpp:744] Ignoring source layer conv13_mbox_conf
I0218 12:09:09.081063 27426 net.cpp:744] Ignoring source layer conv14_2_mbox_conf
I0218 12:09:09.081077 27426 net.cpp:744] Ignoring source layer conv15_2_mbox_conf
I0218 12:09:09.081089 27426 net.cpp:744] Ignoring source layer conv16_2_mbox_conf
I0218 12:09:09.081097 27426 net.cpp:744] Ignoring source layer conv17_2_mbox_conf
I0218 12:09:09.081101 27426 net.cpp:744] Ignoring source layer mbox_loss
I0218 12:09:09.081256 27426 caffe.cpp:255] Starting Optimization
I0218 12:09:09.081264 27426 solver.cpp:302] Solving MobileNet-SSD
I0218 12:09:09.081267 27426 solver.cpp:303] Learning Rate Policy: multistep
I0218 12:09:09.084375 27426 blocking_queue.cpp:50] Data layer prefetch queue empty

是在尝试训练MobileNet-SSd遇到的错误，有人知道为啥吗？

Reply to 求助！！！！！！caffe的错误 Data layer prefetch queue empty on Mon, 18 Feb 2019 04:56:42 GMT

刘看山 — Mon, 18 Feb 2019 04:56:42 GMT

Reply to 求助！！！！！！caffe的错误 Data layer prefetch queue empty on Mon, 18 Feb 2019 10:14:04 GMT

刘看山 — Mon, 18 Feb 2019 10:14:04 GMT

问题已经解决了，这个问题是由于一个check函数造成的，之前由于报错，我uncomment了，导致了错误，现在需要调换二者，解决方案是：

将 match_functions.cpp 里面的这个函数前面几行，改为这个

void caffe_rng_uniform(const int n, Dtype a, Dtype b, Dtype* r) {
  CHECK_GE(n, 0);
  CHECK(r);

  if(a > b) {
    Dtype c = a;
    a = b;
    b = c;
  }
  CHECK_LE(a, b);

最后完美解决：