Running ModelPack on EVK

In this section, we will explain how to run ModelPack on the i.MX8M Plus EVK using Deep View VisionPack. Deep View VisionPack allows developers to deploy AI Vision Models created with eIQ Toolkit to any NXP i.MX 8M device or development platform with ease. By using available hardware acceleration of the SOC, the VisionPack provides you with a highly configurable, fully optimized, tested AI Vision pipeline right out of the box.

Prerequisites

1. i.MX 8M Plus EVK (Maivin Guide Coming Soon!)
2. NXP Yocto BSP 5.10.72 or newer. 
   • Refer to the loading Linux on SD card article for how to get the latest Linux BSP on your EVK.
3. Camera connected to EVK - MINISASTOCSI
4. HDMI Display connected to the EVK

Installation 

If you don't have a valid Deep View VisionPack already installed on your EVK, please install that first by referring to the VisionPack QuickStart guide.

Running the model

Before running the samples, we need to copy the model file to the EVK first.  Use a SCP tool to transfer the model file (.rtm typically) to the target.  On windows, open a cmd window, change to the directory the model is stored and use the SCP command to transfer the file:

# scp mpk-cards.rtm user@evk.address:/home/root/
e.g. 
# scp mpk-cards.rtm root@10.10.40.209:/home/root/

Once the model has been transferred to the EVK and the visual interface of EVK operating system is displayed on the screen, we just need to run the following on the EVK:

Connect to your EVK via SSH and source VisionPack script.

# source visionpack-latest/bin/env.sh

If this command was previously appended to the .bashrc you can skip to source the library. At this point we are ready to start running our model.  The following will start a g-streamer video stream and show the model output (bounding boxes) on the screen when detecting the object.  Refer to the VisionPack user guides for more command options and details.

# sh gst-detect.sh <ModelName.rtm>
e.g.
# sh gst-detect.sh mpk-cards.rtm

If every step was successfully executed, you should see the following information inside the command line or ssh client.

The camera feed should display on the screen, and start detecting cards.