1. Gather required equipment.

      • A PC with:
        • Linux (Ubuntu 16.04, 64-bit version) operating system*
        • USB 3.0 port preferable
        • Internet connection
      • Dev Hardware
        • Plai™ Plug 2801
        • Plai™ Plug 2803
        • SolidRun’s Mini SOM
      • GTI SDK source

*Also supports Windows and Android OS. Refer to SDK User Guide for details.

  • 2801
  • 2803

 2. Unpack the SDK source.

1. Open terminal and unpack the source:

tar zxvf GTISDK-Linux_x86_64-v4.0.0.tgz

2. Navigate to the SDK root directory:

cd GTISDK-Linux_x86_64-v4.0.0

4. Run this command in the terminal to detect missing dependencies, drivers, and so on:

source SourceMe.env

 3. Run demos.

1. Insert Plai Plug into the USB port.
2. Run the following instructions (Python or C) to start the four application demos: Images, Video, Camera (USB camera required) or Slideshow.

FOR PYTHON USERS:

1. Install GTI Library binding:

pip install --user ./Python/Lib/gtilib/

2. Navigate to the Python directory:

cd Apps/Python

3. Perform inference on one single image:

  python demo.py image [model file] [image filename]

python demo.py image ../Models/2801/gti_gnet3_fc20_2801.model

../Data/Image_bmp_c20/bridges.jpg

Or, perform inference on one video stream:

  python demo.py video [model file]

python demo.py video ../Models/2801/gti_gnet3_fc20_2801.model

../Data/Image_mp4/video_20class.mp4

Or, perform inference on images from the on-board camera:

  python demo.py camera [model file] 0

python demo.py camera ../Models/2801/gti_gnet3_fc20_2801.model 0

Or, perform inference on images from a directory:

  python demo.py slideshow [model file] [directory name]

python demo.py slideshow ../Models/2801/gti_gnet3_fc20_2801.model ../Data/Image_bmp_c20/

FOR C USERS:

Navigate to the application directory:

cd Apps/Demo

Perform inference on one single image:

&nbsp demo image [model file] [image filename]

demo image ../Models/2801/gti_gnet3_fc20_2801.model ../Data/Image_bmp_c20/bridges.jpg

&nbsp Or, perform inference on one video stream:

&nbsp demo video [model file]

demo video ../Models/2801/gti_gnet3_fc20_2801.model

../Data/Image_mp4/video_20class.mp4

&nbsp
Or, perform inference on images from the on-board camera:

&nbsp demo camera [model file] 0

demo camera ../Models/2801/gti_gnet3_fc20_2801.model 0

&nbsp
Or, perform inference on images from a directory:

&nbsp demo slideshow [model file] [directory name]

demo slideshow ../Models/2801/gti_gnet3_fc20_2801.model ../Data/Image_bmp_c20/

 2. Unpack the SDK source.

1. Open terminal and unpack the source:

tar zxvf GTISDK-Linux_x86_64-v4.0.0.tgz

2. Navigate to the SDK root directory:

cd GTISDK-Linux_x86_64-v4.0.0

4. Run this command in the terminal to detect missing dependencies, drivers, and so on:

source SourceMe.env

 3. Run demos.

1. Insert Plai Plug into the USB port.
2. Run the following instructions (Python or C) to start the four application demos: Images, Video, Camera (USB camera required) or Slideshow.

FOR PYTHON USERS:

1. Install GTI Library binding:

pip install --user ./Python/Lib/gtilib/

2. Navigate to the Python directory:

cd Apps/Python

3. Perform inference on one single image:

  python demo.py image [model file] [image filename]

python demo.py image ../Models/2803/gti_gnet1_fc1000_2803.model

../Data/Image_bmp_c1000/beach.bmp

Or, perform inference on one video stream:

  python demo.py video [model file]

python demo.py video ../Models/2803/gti_mnet_fc40_2803.model

../Data/Image_mp4/video_40class.avi

Or, perform inference on images from the on-board camera:

  python demo.py camera [model file] 0

python demo.py camera ../Models/2803/gti_gnet1_fc1000_2803.model 0

Or, perform inference on images from a directory:

  python demo.py slideshow [model file] [directory name]

python demo.py slideshow ../Models/2803/gti_gnet1_fc1000_2803.model ../Data/Image_bmp_c1000/

FOR C USERS:

Navigate to the application directory:

cd Apps/Demo

Perform inference on one single image:

demo image [model file] [image filename]

demo image ../Models/2803/gti_gnet1_fc1000_2803.model ../Data/Image_bmp_c1000/beach.bmp

Or, perform inference on one video stream:

demo video [model file]

demo video ../Models/2803/gti_mnet_fc40_2803.model

../Data/Image_mp4/video_40class.avi

Or, perform inference on images from the on-board camera:

demo camera [model file] 0

demo camera ../Models/2803/gti_gnet1_fc1000_2803.model 0

Or, perform inference on images from a directory:

demo slideshow [model file] [directory name]

demo slideshow ../Models/2803/gti_gnet1_fc1000_2803.model ../Data/Image_bmp_c1000/
Congratulations! You have now tested and experienced the power of the Lightspeeur® AI accelerator.  Learn more about the SDK or how to convert and deploy your model onto the GTI accelerator.