ddef SKConv(input, M, r, L=32, stride=1, is_training=True):
input_feature = input.get_shape().as_list()[3]
d = max(int(input_feature / r), L)
net = input
with slim.arg_scope([slim.conv2d, slim.fully_connected], activation_fn=tf.nn.relu):
for i in range(M):
net = slim.conv2d(net, input_feature, [3+i*2, 3+i*2], rate=1+i, stride=stride)
net = slim.batch_norm(net, decay=0.9, center=True, scale=True, epsilon=1e-5,
updates_collections=tf.GraphKeys.UPDATE_OPS, is_training=is_training)
net = tf.nn.relu(net)
if i == 0:
fea_U = net
else:
fea_U = tf.add(fea_U, net)
gap = tflearn.global_avg_pool(fea_U)
fc = slim.fully_connected(gap, d, activation_fn=None)
fcs = fc
for _ in range(M):
fcs = slim.fully_connected(fcs, input_feature, activation_fn=None)
if _ == 0:
att_vec = fcs
else:
att_vec = tf.add(att_vec, fcs)
att_vec = tf.expand_dims(att_vec, axis=1)
att_vec = tf.expand_dims(att_vec, axis=1)
att_vec_softmax = tf.nn.softmax(att_vec)
fea_v = tf.multiply(fea_U, att_vec_softmax)
return fea_v
这是tensroflow版本的，在keras上运行会报错“'tuple' object has no attribute 'layer'”，有没有人可以改写一下或者有其他keras实现的版本？
