diff --git a/pyzoo/zoo/examples/streaming/objectdetection/README.md b/pyzoo/zoo/examples/streaming/objectdetection/README.md
index d7d18b4cc26..6a2a953d40d 100644
--- a/pyzoo/zoo/examples/streaming/objectdetection/README.md
+++ b/pyzoo/zoo/examples/streaming/objectdetection/README.md
@@ -15,6 +15,8 @@ So, there are two applications in this example: image_path_writer and streaming_
 ## Run this example
 Make sure all nodes can access image files, model and text files. Local file system/HDFS/Amazon S3 are supported.
 
+Pls ensure all paths exist and accessible, and `streaming_path` is empty. Note that `streaming_object_detection` and `image_path_writer` should use the same `streaming_path`.
+
 1. Start streaming_object_detection
 ```
 MASTER=...
diff --git a/pyzoo/zoo/examples/streaming/objectdetection/image_path_writer.py b/pyzoo/zoo/examples/streaming/objectdetection/image_path_writer.py
index 2c8aa3de569..6e888f3dd9d 100644
--- a/pyzoo/zoo/examples/streaming/objectdetection/image_path_writer.py
+++ b/pyzoo/zoo/examples/streaming/objectdetection/image_path_writer.py
@@ -19,6 +19,7 @@
 from time import sleep
 from os import listdir, rename, mkdir, remove
 from os.path import isfile, join
+import shutil
 
 
 def package_path_to_text(streaming_path, file_path, batch=10, delay=3):
@@ -49,8 +50,8 @@ def package_path_to_text(streaming_path, file_path, batch=10, delay=3):
         with open(join(tmpDir, str(index) + ".txt"), "w") as text_file:
             text_file.writelines(files[curr:last])
         # Move to streaming location
-        rename(text_file.name,
-               batch_file_name)
+        shutil.move(text_file.name,
+                    batch_file_name)
         print("Writing to " + batch_file_name)
         index += 1
         curr = last
diff --git a/zoo/src/main/scala/com/intel/analytics/zoo/common/Utils.scala b/zoo/src/main/scala/com/intel/analytics/zoo/common/Utils.scala
index 0416befb204..9a977eec959 100644
--- a/zoo/src/main/scala/com/intel/analytics/zoo/common/Utils.scala
+++ b/zoo/src/main/scala/com/intel/analytics/zoo/common/Utils.scala
@@ -18,12 +18,12 @@ package com.intel.analytics.zoo.common
 
 import java.io._
 import java.nio.file.attribute.PosixFilePermissions
+import java.nio.file.{Path => JPath}
 
-import com.intel.analytics.bigdl.utils.File
 import org.apache.commons.io.filefilter.WildcardFileFilter
 import org.apache.hadoop.conf.Configuration
-import org.apache.hadoop.fs.{FileSystem, Path}
-import java.nio.file.{Path => JPath}
+import org.apache.hadoop.fs.{FSDataInputStream, FSDataOutputStream, FileSystem, Path}
+import org.apache.hadoop.io.IOUtils
 import org.apache.log4j.Logger
 
 import scala.collection.mutable.ArrayBuffer
@@ -94,7 +94,18 @@ private[zoo] object Utils {
    * @return Array[Byte]
    */
   def readBytes(path: String): Array[Byte] = {
-    File.readBytes(path)
+    var fs: FileSystem = null
+    var in: FSDataInputStream = null
+    try {
+      fs = getFileSystem(path)
+      in = fs.open(new Path(path))
+      val byteArrayOut = new ByteArrayOutputStream()
+      IOUtils.copyBytes(in, byteArrayOut, 1024, true)
+      byteArrayOut.toByteArray
+    } finally {
+      if (null != in) in.close()
+      if (null != fs) fs.close()
+    }
   }
 
   /**
@@ -132,7 +143,7 @@ private[zoo] object Utils {
    *                    get from cache (may shared with our connections)
    * @return hadoop.fs.FileSystem
    */
-  def getFileSystem(fileName: String, newInstance: Boolean = false): FileSystem = {
+  def getFileSystem(fileName: String, newInstance: Boolean = true): FileSystem = {
     if (newInstance) {
       FileSystem.newInstance(new Path(fileName).toUri, new Configuration())
     } else {
@@ -183,7 +194,16 @@ private[zoo] object Utils {
    * @param isOverwrite Overwrite exiting file or not
    */
   def saveBytes(bytes: Array[Byte], fileName: String, isOverwrite: Boolean = false): Unit = {
-    File.saveBytes(bytes, fileName, isOverwrite)
+    var fs: FileSystem = null
+    var out: FSDataOutputStream = null
+    try {
+      fs = getFileSystem(fileName)
+      out = fs.create(new Path(fileName), isOverwrite)
+      IOUtils.copyBytes(new ByteArrayInputStream(bytes), out, 1024, true)
+    } finally {
+      if (null != out) out.close()
+      if (null != fs) fs.close()
+    }
   }
 
   def logUsageErrorAndThrowException(errMessage: String, cause: Throwable = null): Unit = {
diff --git a/zoo/src/main/scala/com/intel/analytics/zoo/examples/streaming/objectdetection/README.md b/zoo/src/main/scala/com/intel/analytics/zoo/examples/streaming/objectdetection/README.md
index 5f8ee2fc42b..6d1c85b3141 100644
--- a/zoo/src/main/scala/com/intel/analytics/zoo/examples/streaming/objectdetection/README.md
+++ b/zoo/src/main/scala/com/intel/analytics/zoo/examples/streaming/objectdetection/README.md
@@ -14,6 +14,8 @@ So, there are two applications in this example: ImagePathWriter and StreamingObj
 ## Run this example
 Make sure all nodes can access image files, model and text files. Local file system/HDFS/Amazon S3 are supported.
 
+Pls ensure all paths exist and accessible, and `streamingPath` is empty. Note that `StreamingObjectDetection` and `ImagePathWriter` should use the same `streamingPath`.
+
 1. Start StreamingObjectDetection
 ```
 MASTER=...