support image target resize

cnn_e2e/of_cnn_train_val.py

@@ -97,6 +97,7 @@ def main():
                             save_summary_steps=num_val_steps, batch_size=val_batch_size)
             for i in range(num_val_steps):
                 InferenceNet().async_get(metric.metric_cb(epoch, i))
+        snapshot.save('epoch_{}'.format(epoch))
 
 
 if __name__ == "__main__":

cnn_e2e/ofrecord_util.py

@@ -46,7 +46,7 @@ def load_imagenet_for_training(args):
         flow.data.ImageResizePreprocessor(args.image_size, args.image_size),
         flow.data.ImagePreprocessor('mirror'),
     ])
-    return load_imagenet(args, train_batch_size, args.train_data_dir, args.train_data_part_num, 
+    return load_imagenet(args, train_batch_size, args.train_data_dir, args.train_data_part_num,
                          codec)
 
 
@@ -55,7 +55,9 @@ def load_imagenet_for_validation(args):
     val_batch_size = total_device_num * args.val_batch_size_per_device
     codec=flow.data.ImageCodec(
         [
-            flow.data.ImageResizePreprocessor(args.image_size, args.image_size),
+            flow.data.ImageTargetResizePreprocessor(resize_shorter=256),
+            flow.data.ImageCenterCropPreprocessor(args.image_size, args.image_size),
+            #flow.data.ImageResizePreprocessor(args.image_size, args.image_size),
         ]
     )
     return load_imagenet(args, val_batch_size, args.val_data_dir, args.val_data_part_num, codec)
@@ -75,4 +77,4 @@ def load_synthetic(args):
         shape=(args.image_size, args.image_size, 3), dtype=flow.float, batch_size=batch_size
     )
 
-    return label, image
\ No newline at end of file
+    return label, image

cnn_e2e/resnet_model.py

@@ -22,7 +22,7 @@ def _conv2d(
     trainable=True,
     weight_initializer=flow.variance_scaling_initializer(data_format="NCHW"),
     #weight_initializer=flow.variance_scaling_initializer(3, 'fan_in', 'random_normal', data_format="NCHW"),
-    weight_regularizer=flow.regularizers.l2(1e-4),
+    weight_regularizer=flow.regularizers.l2(1.0/32768),
 ):
     weight = flow.get_variable(
         name + "-weight",
@@ -143,7 +143,7 @@ def resnet50(images, trainable=True):
             #kernel_initializer=flow.variance_scaling_initializer(3, 'fan_in', 'random_normal'),
             kernel_initializer=flow.xavier_uniform_initializer(),
             bias_initializer=flow.zeros_initializer(),
-            kernel_regularizer=flow.regularizers.l2(1e-4),
+            kernel_regularizer=flow.regularizers.l2(1.0/32768),
             trainable=trainable,
             name="fc1001",
         )