fix errors

yanx27 · May 20, 2020 · 5c28541 · 5c28541
1 parent ecd6acc
commit 5c28541
Show file tree

Hide file tree

Showing 4 changed files with 7 additions and 7 deletions.
diff --git a/README.md b/README.md
@@ -27,7 +27,7 @@ git clone https://github.com/yanx27/PointASNL.git
 Installation instructions for Ubuntu 16.04 (available at **CUDA10**):
 
 * Make sure <a href="https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html">CUDA</a>  and <a href="https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html">cuDNN</a> are installed. Only this configurations has been tested: 
-     - Python 3.6.9, TensorFlow 1.13.0, CUDA 10.1
+     - Python 3.6.9, TensorFlow 1.13.1, CUDA 10.1
 
 
 * Follow <a href="https://www.tensorflow.org/install/pip">Tensorflow installation procedure</a>.

diff --git a/models/pointasnl_sem_seg.py b/models/pointasnl_sem_seg.py
@@ -50,7 +50,7 @@ def get_model(point_cloud, is_training, num_class, bn_decay=None, weight_decay=N
     return net, end_points
 
 
-def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_decay=1e-4):
+def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_decay=1e-4, radius=0.07):
     """
     pred: BxNxC,
     label: BxN,
@@ -59,7 +59,7 @@ def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_dec
     regularization_losses = [tf.nn.l2_loss(v) for v in tf.global_variables() if 'weights' in v.name]
     regularization_loss = weights_decay * tf.add_n(regularization_losses)
     classify_loss = tf.losses.sparse_softmax_cross_entropy(labels=label, logits=pred, weights=smpw)
-    uniform_loss = get_repulsion_loss(end_points['l1_xyz'], nsample=20, radius=0.07)
+    uniform_loss = get_repulsion_loss(end_points['l1_xyz'], nsample=20, radius=radius)
     weight_reg = tf.add_n(tf.get_collection('losses'))
     classify_loss_mean = tf.reduce_mean(classify_loss, name='classify_loss_mean')
     total_loss = classify_loss_mean + weight_reg + uniform_weight * uniform_loss + regularization_loss

diff --git a/models/pointasnl_sem_seg_res.py b/models/pointasnl_sem_seg_res.py
@@ -68,7 +68,7 @@ def get_model(point_cloud, is_training, num_class, bn_decay=None, weight_decay=N
     return net, end_points
 
 
-def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_decay=1e-4):
+def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_decay=1e-4, radius=0.07):
     """
     pred: BxNxC,
     label: BxN,
@@ -77,7 +77,7 @@ def get_loss(pred, label, end_points, smpw=1.0, uniform_weight=0.01, weights_dec
     regularization_losses = [tf.nn.l2_loss(v) for v in tf.global_variables() if 'weights' in v.name]
     regularization_loss = weights_decay * tf.add_n(regularization_losses)
     classify_loss = tf.losses.sparse_softmax_cross_entropy(labels=label, logits=pred, weights=smpw)
-    uniform_loss = get_repulsion_loss(end_points['l1_xyz'], nsample=20, radius=0.07)
+    uniform_loss = get_repulsion_loss(end_points['l1_xyz'], nsample=20, radius=radius)
     weight_reg = tf.add_n(tf.get_collection('losses'))
     classify_loss_mean = tf.reduce_mean(classify_loss, name='classify_loss_mean')
     total_loss = classify_loss_mean + weight_reg + uniform_weight * uniform_loss + regularization_loss

diff --git a/utils/pointasnl_util.py b/utils/pointasnl_util.py
@@ -147,12 +147,12 @@ def SampleWeights(new_point, grouped_xyz, mlps, is_training, bn_decay, weight_de
         new_group_features = tf.reshape(new_group_features, (batch_size, npoint, nsample, channel))
         for i, c in enumerate(mlps):
             activation = tf.nn.relu if i < len(mlps) - 1 else None
-            new_group_weights = tf_util.conv2d(new_group_features, c, [1, 1],
+            new_group_features = tf_util.conv2d(new_group_features, c, [1, 1],
                                                padding='VALID', stride=[1, 1],
                                                bn=bn, is_training=is_training,
                                                scope='mlp2_%d' % (i), bn_decay=bn_decay, weight_decay=weight_decay,
                                                activation_fn=activation)
-        new_group_weights = tf.nn.softmax(new_group_weights, axis=2)  # (batch_size, npoint,nsample, mlp[-1)
+        new_group_weights = tf.nn.softmax(new_group_features, axis=2)  # (batch_size, npoint,nsample, mlp[-1)
         return new_group_weights
 
 def AdaptiveSampling(group_xyz, group_feature, num_neighbor, is_training, bn_decay, weight_decay, scope, bn):