MTCNN-Tensorflow gen_hard_example出错：Key fc1/alphas not found in checkpoint

@AITTSMD 您好，感谢您的开源工作。想通过您的工作熟悉一下MTCNN的训练过程，我已经按照您的指引以默认配置完成了PNet的训练，接下来是按照【After training PNet, run gen_hard_example to generate training data(Face Detection Part) for RNet.】的指引，运行gen_hard_example脚本生成RNet的训练数据，但是总是出错，在加载预训练模型RNet-14的model时出错： ../data/MTCNN_model/RNet_landmark/RNet-14 restore models' param 2017-11-01 16:41:20.483995: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key fc1/alphas not found in checkpoint 2017-11-01 16:41:20.484165: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key fc1/alphas not found in checkpoint 请教该如何解决？

Nov 01 '17 00:11 ddxu

@ddxu 您好，注意检查您的tensorflow的版本，我用的是1.2.1

Nov 01 '17 01:11 AITTSMD

使用的版本的是1.3.0 遇到了同样的问题，看来只能降版本了

Nov 01 '17 06:11 ghost

@ddxu it should be solved by: https://github.com/AITTSMD/MTCNN-Tensorflow/commit/519b453975a8dbdc7b0eecdb4f8735b7b6c80d46

Nov 01 '17 08:11 ilyanelken

把楼上的那次commit还原之后，去掉最后那个参数，竟然可以正常运行了...

Nov 01 '17 08:11 ghost

按照楼上的去掉最后那个参数 #fc1 = slim.fully_connected(fc_flatten, num_outputs=256,scope="fc1", activation_fn=prelu) 改成 fc1 = slim.fully_connected(fc_flatten, num_outputs=256,scope="fc1")就可以了！

Nov 03 '17 12:11 94happy

@ilyanelken Hi, can you specify where you find the original author needs some "special" relu act function for Rnet and Onet? I check the paper and author's code?

Nov 24 '17 07:11 flankechen

@ilyanelken and I think that prelu act function is already as default in the arg scope.

Nov 24 '17 07:11 flankechen

@flankechen Hi, you can see the original code that was written by the author's: https://github.com/kpzhang93/MTCNN_face_detection_alignment/blob/master/code/codes/MTCNNv1/model/det2.prototxt There is PReLU. The arg scope is valid for conv2d ops only and not fully_connected

Jan 22 '18 21:01 ilyanelken

what are you talking about? I faced the same problem, anyone who knows the solution?please give a hand