YOLO object detector for Movidius Neural Compute Stick (NCS)
Clone or download
Pull request Compare This branch is 4 commits ahead of gudovskiy:master.
Latest commit dfd6735 Apr 25, 2018

README.md

YOLO for PYNQ-Z1 and Intel/Movidius Neural Compute Stick (NCS)

This project is derived from yoloNCS and is intended to be used on the PYNQ-Z1 board.

alt text

Using this repo on your PYNQ-Z1

To use this code on your PYNQ-Z1, just follow these steps:

  1. Install NCSDK in API-mode on your PYNQ-Z1 as explained here: Setting up the PYNQ-Z1 for the Intel Movidius NCS

  2. Clone this repo onto your PYNQ-Z1 in this directory: /home/xilinx/jupyter_notebooks

  3. Boot the PYNQ-Z1, open Jupyter in a web browser (http://pynq:9090) and open one of the notebooks

News

  • Camera App is working.
  • YOLOv1 Tiny is working.

Protobuf Model files

./prototxt/

Download Pretrained Caffe Models to ./weights/

Compilation

  • Compile .prototxt and corresponding .caffemodel (with the same name) to get NCS graph file. For example: "mvNCCompile prototxt/yolo_tiny_deploy.prototxt -w weights/yolo_tiny_deploy.caffemodel -s 12"
  • The compiled binary file "graph" has to be in main folder after this step.

Single Image Script

  • Run "yolo_example.py" to process a single image. For example: "python3 py_examples/yolo_example.py images/dog.jpg" to get detections as below.

Camera Input Script

  • Run "object_detection_app.py" to process a videos from your camera. For example: "python3 py_examples/object_detection_app.py" to get camera detections as below.
  • Modify script arguments if needed.
  • Press "q" to exit app.