78666621 commented 3 years ago

cified using deprecated input fields. W0123 22:15:11.515242 20449 upgrade_proto.cpp:71] Note that future Caffe releases will only support input layers and not input fields. I0123 22:15:11.531896 20449 net.cpp:762] Ignoring source layer pool5 I0123 22:15:11.531915 20449 net.cpp:762] Ignoring source layer fc6 I0123 22:15:11.531919 20449 net.cpp:762] Ignoring source layer relu6 I0123 22:15:11.531922 20449 net.cpp:762] Ignoring source layer drop6 I0123 22:15:11.531925 20449 net.cpp:762] Ignoring source layer fc7 I0123 22:15:11.531931 20449 net.cpp:762] Ignoring source layer relu7 I0123 22:15:11.531935 20449 net.cpp:762] Ignoring source layer drop7 I0123 22:15:11.531939 20449 net.cpp:762] Ignoring source layer fc8_pascal I0123 22:15:11.539551 20449 solver.cpp:337] Iteration 0, Testing net (#0) F0123 22:15:12.482286 20449 syncedmem.cpp:56] Check failed: error == cudaSuccess (2 vs. 0) out of memory Check failure stack trace: 以上为报错内容，是因为电脑的gup性能太差导致的吗？台式机的算力为3，batch_size设置为1读取图片，也不行，打扰您了，期望得到您的回复。

yun-liu commented 3 years ago

台式机的算力为3是什么意思？您用的是什么GPU？

另外，我看您的这个输出，如果您是在测试的话，是不是用错caffe model了？

78666621 commented 3 years ago

谢谢您的回复，已经换了一台电脑，已经正常可以训练了，其次还有一个问题打扰您一下 image_data_param { root_folder: "../../data/" source: "../../data/bsds_pascal_train_pair.lst" batch_size: 1 shuffle: true new_height: 0 new_width: 0 } } layer { name: "data" type: "ImageLabelmapData" top: "data" top: "label" include { phase: TEST } transform_param { mirror: false mean_value: 104.00699 mean_value: 116.66877 mean_value: 122.67892 } image_data_param { root_folder: "../../data/" source: "../../data/bsds_pascal_train_pair.lst"

Just setup the network. No real online testing

batch_size: 1
shuffle: true
new_height: 0
new_width: 0

} } 我注意到这段代码里面分了训练集和测试集，然后您备注说 #Just setup the network. No real online testing，是否意味着这个test层没有意义，可以写与前面代码中的训练层一样的路径，实际不会在这里测试数据。期待您的回复。

yun-liu commented 3 years ago

是的，这个test层没有意义

yun-liu / RCF

用rcf测试发生的错误 #123

Just setup the network. No real online testing