fix the std bug

leads to unstable testing

classification_ModelNet40/main.py

@@ -1,3 +1,7 @@
+"""
+Usage:
+python main.py --model PointMLP --msg demo
+"""
 import argparse
 import os
 import logging
@@ -28,6 +32,7 @@ def parse_args():
     parser.add_argument('--epoch', default=300, type=int, help='number of epoch in training')
     parser.add_argument('--num_points', type=int, default=1024, help='Point Number')
     parser.add_argument('--learning_rate', default=0.1, type=float, help='learning rate in training')
+    parser.add_argument('--min_lr', default=0.005, type=float, help='min lr')
     parser.add_argument('--weight_decay', type=float, default=2e-4, help='decay rate')
     parser.add_argument('--seed', type=int, help='random seed')
     parser.add_argument('--workers', default=8, type=int, help='workers')
@@ -122,7 +127,7 @@ def main():
     optimizer = torch.optim.SGD(net.parameters(), lr=args.learning_rate, momentum=0.9, weight_decay=args.weight_decay)
     if optimizer_dict is not None:
         optimizer.load_state_dict(optimizer_dict)
-    scheduler = CosineAnnealingLR(optimizer, args.epoch, eta_min=1e-3, last_epoch=start_epoch - 1)
+    scheduler = CosineAnnealingLR(optimizer, args.epoch, eta_min=args.min_lr, last_epoch=start_epoch - 1)
 
     for epoch in range(start_epoch, args.epoch):
         printf('Epoch(%d/%s) Learning Rate %s:' % (epoch + 1, args.epoch, optimizer.param_groups[0]['lr']))

classification_ModelNet40/models/pointmlp.py

@@ -170,11 +170,11 @@ class LocalGrouper(nn.Module):
             grouped_points = torch.cat([grouped_points, grouped_xyz],dim=-1)  # [B, npoint, k, d+3]
         if self.normalize is not None:
             if self.normalize =="center":
-                std, mean = torch.std_mean(grouped_points, dim=2, keepdim=True)
+                mean = torch.mean(grouped_points, dim=2, keepdim=True)
             if self.normalize =="anchor":
                 mean = torch.cat([new_points, new_xyz],dim=-1) if self.use_xyz else new_points
                 mean = mean.unsqueeze(dim=-2)  # [B, npoint, 1, d+3]
-                std = torch.std(grouped_points-mean)
+            std = torch.std((grouped_points-mean).reshape(B,-1),dim=-1,keepdim=True).unsqueeze(dim=-1).unsqueeze(dim=-1)
             grouped_points = (grouped_points-mean)/(std + 1e-5)
             grouped_points = self.affine_alpha*grouped_points + self.affine_beta
 

classification_ScanObjectNN/models/pointmlp.py

@@ -170,11 +170,11 @@ class LocalGrouper(nn.Module):
             grouped_points = torch.cat([grouped_points, grouped_xyz],dim=-1)  # [B, npoint, k, d+3]
         if self.normalize is not None:
             if self.normalize =="center":
-                std, mean = torch.std_mean(grouped_points, dim=2, keepdim=True)
+                mean = torch.mean(grouped_points, dim=2, keepdim=True)
             if self.normalize =="anchor":
                 mean = torch.cat([new_points, new_xyz],dim=-1) if self.use_xyz else new_points
                 mean = mean.unsqueeze(dim=-2)  # [B, npoint, 1, d+3]
-                std = torch.std(grouped_points-mean)
+            std = torch.std((grouped_points-mean).reshape(B,-1),dim=-1,keepdim=True).unsqueeze(dim=-1).unsqueeze(dim=-1)
             grouped_points = (grouped_points-mean)/(std + 1e-5)
             grouped_points = self.affine_alpha*grouped_points + self.affine_beta
 

part_segmentation/models/pointMLP.py

@@ -169,11 +169,11 @@ class LocalGrouper(nn.Module):
             grouped_points = torch.cat([grouped_points, grouped_xyz],dim=-1)  # [B, npoint, k, d+3]
         if self.normalize is not None:
             if self.normalize =="center":
-                std, mean = torch.std_mean(grouped_points, dim=2, keepdim=True)
+                mean = torch.mean(grouped_points, dim=2, keepdim=True)
             if self.normalize =="anchor":
                 mean = torch.cat([new_points, new_xyz],dim=-1) if self.use_xyz else new_points
-                mean = mean.unsqueeze(dim=-2)  # [B, npoint, 1, d+3]
-                std = torch.std(grouped_points-mean)
+                mean = mean.unsqueeze(dim=-2)  # [B, npoint, 1, d+3]	
+            std = torch.std((grouped_points-mean).reshape(B,-1),dim=-1,keepdim=True).unsqueeze(dim=-1).unsqueeze(dim=-1)
             grouped_points = (grouped_points-mean)/(std + 1e-5)
             grouped_points = self.affine_alpha*grouped_points + self.affine_beta