Pierre Fenoll 1f121dd3eb CPU inference on desktop Linux with PyTorch v1.4.0

Signed-off-by: Pierre Fenoll <pierrefenoll@gmail.com>

2020-10-17 21:03:22 +02:00

2.6 KiB

Raw Blame History

layout	title	parent	nav_order
default	Object Classification	Solutions	TODO

MediaPipe Object Classification

{: .no_toc }

TOC {:toc}

Example Apps

Note: To visualize a graph, copy the graph and paste it into MediaPipe Visualizer. For more information on how to visualize its associated subgraphs, please see visualizer documentation.

Desktop

Live Camera Input

Please first see general instructions for desktop on how to build MediaPipe examples.

Video File Input

With a PyTorch Model

This uses a MobileNetv2 trace model from PyTorch Hub. To fetch and prepare it, run:

python mediapipe/models/trace_mobilenetv2.py

The pipeline is implemented in this graph.

To build the application, run:

bazel build -c opt --define MEDIAPIPE_DISABLE_GPU=1 mediapipe/examples/desktop/object_classification:object_classification_pytorch_cpu

To run the application, replace <input video path> and <output video path> in the command below with your own paths:

Tip: You can find a test video available in mediapipe/examples/desktop/object_detection.

GLOG_logtostderr=1 bazel-bin/mediapipe/examples/desktop/object_classification/object_classification_pytorch_cpu \
  --calculator_graph_config_file=mediapipe/graphs/object_classification/object_classification_desktop_live.pbtxt \
  --input_side_packets=input_video_path=<input video path>,output_video_path=<output video path>

2.6 KiB Raw Blame History