diff --git a/python/python/jetson/utils/__init__.py b/python/python/jetson/utils/__init__.py
index 2a938316..8a753647 100644
--- a/python/python/jetson/utils/__init__.py
+++ b/python/python/jetson/utils/__init__.py
@@ -3,4 +3,6 @@
 
 from jetson_utils_python import *
 
+from .video_loader import *
+
 VERSION = '1.0.0'
diff --git a/python/python/jetson/utils/video_loader.py b/python/python/jetson/utils/video_loader.py
new file mode 100644
index 00000000..d19b0c70
--- /dev/null
+++ b/python/python/jetson/utils/video_loader.py
@@ -0,0 +1,155 @@
+#
+# Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a
+# copy of this software and associated documentation files (the "Software"),
+# to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense,
+# and/or sell copies of the Software, and to permit persons to whom the
+# Software is furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+# DEALINGS IN THE SOFTWARE.
+#
+
+import queue
+
+import gi
+gi.require_version("Gst", "1.0")
+from gi.repository import Gst
+
+Gst.init(None)
+
+class VideoLoader:
+
+    def __init__(self, logger, width=1280, height=720, framerate="5/1"):
+        self.__logger = logger
+        self.__width = width
+        self.__height = height
+        self.__framerate = framerate
+
+        self.__pipeline = None
+        self.__queue = queue.Queue()
+
+    def load(self, filename):
+        if self.__pipeline:
+            raise Exception("Already processing a file")
+        self.__pipeline = Gst.Pipeline.new("video-loader")
+
+        bus = self.__pipeline.get_bus()
+        bus.add_signal_watch()
+        bus.connect("message", self.__on_message)
+
+        self.__build_pipeline(filename)
+        self.__pipeline.set_state(Gst.State.PLAYING)
+
+
+    def stop(self):
+        if self.__pipeline:
+            self.__pipeline.set_state(Gst.State.NULL)
+            self.__pipeline = None
+
+    def is_loading(self):
+        return self.__pipeline != None
+
+    def sample_next(self):
+        return self.__queue.get()
+
+    def sample_done(self):
+        self.__queue.task_done()
+
+    def __build_pipeline(self, filename):
+        source = Gst.ElementFactory.make("filesrc")
+        source.set_property("location", filename)
+        self.__pipeline.add(source)
+
+        decodebin = Gst.ElementFactory.make("decodebin")
+        decodebin.connect("pad-added", self.__on_pad_added)
+        self.__pipeline.add(decodebin)
+
+        source.link(decodebin)
+
+    def __on_message(self, bus, message):
+        t = message.type
+        if t == Gst.MessageType.EOS:
+            self.stop()
+        elif t == Gst.MessageType.ERROR:
+            self.stop()
+            err, debug = message.parse_error()
+            self.__logger.error("Error: %s (%s)" % (err, debug))
+
+    def __on_pad_added(self, decodebin, pad):
+        caps_str = pad.query_caps(None).to_string()
+        if caps_str.startswith("video"):
+            self.__logger.info("Detected video stream, processing...")
+        elif caps_str.startswith("audio"):
+            self.__logger.info("Detected audio stream, ignoring.")
+            return
+
+        queue = Gst.ElementFactory.make("queue")
+        self.__pipeline.add(queue)
+
+        # rate first so we dn't process as many frames.
+        videorate = Gst.ElementFactory.make("videorate")
+        self.__pipeline.add(videorate)
+
+        caps = "video/x-raw,framerate=%s" % self.__framerate
+        ratecaps = Gst.ElementFactory.make("capsfilter")
+        ratecaps.set_property("caps", Gst.Caps(caps))
+        self.__pipeline.add(ratecaps)
+
+        # scale image in GPU if needed.
+        videoscale = Gst.ElementFactory.make("nvvidconv")
+        self.__pipeline.add(videoscale)
+
+        caps = "video/x-raw(memory:NVMM),format=NV12"
+        if self.__width > 0:
+            caps += ",width=%d" % self.__width
+        if self.__height > 0:
+            caps += ",height=%d" % self.__height
+
+        scalecaps = Gst.ElementFactory.make("capsfilter")
+        scalecaps.set_property("caps", Gst.Caps(caps))
+        self.__pipeline.add(scalecaps)
+
+        # copy to main memory.
+        videoconvert = Gst.ElementFactory.make("nvvidconv")
+        self.__pipeline.add(videoconvert)
+
+        convcaps = Gst.ElementFactory.make("capsfilter")
+        convcaps.set_property("caps", Gst.Caps("video/x-raw"))
+        self.__pipeline.add(convcaps)
+
+        # finally, add the sample sink.
+        appsink = Gst.ElementFactory.make("appsink")
+        appsink.set_property("emit-signals", True)
+        appsink.set_property("max-buffers", 10)
+        appsink.set_property("drop", True)
+        appsink.set_property("sync", True)
+        appsink.connect("new-sample", self.__on_new_sample)
+        self.__pipeline.add(appsink)
+
+        # link everything together.
+        sinkpad = queue.get_static_pad("sink")
+        pad.link(sinkpad)
+
+        queue.link(videorate)
+        videorate.link(ratecaps)
+        ratecaps.link(videoscale)
+        videoscale.link(scalecaps)
+        scalecaps.link(videoconvert)
+        videoconvert.link(convcaps)
+        convcaps.link(appsink)
+
+    def __on_new_sample(self, appsink):
+        sample = appsink.emit("pull-sample")
+        self.__queue.put(sample)
+        return Gst.FlowReturn.OK