Pose estimation processing block

This implements a pose estimation processing block based on PoseNet in Edge Impulse. Use this block to turn raw images into pose vectors, then pair it with an ML block to detect what a person is doing.

How to run this block (locally)

1. Docker:

   1. Build the container:

      $ docker build -t edge-detection-block .
      

   2. Run the block:

      $ docker run -p 4449:4449 -it --rm  edge-detection-block
      

2. Install ngrok and open up port 4449 to the world:

   $ ngrok http 4449
   

   Note down the 'Forwarding' address that starts with https, e.g.:

   Forwarding                    https://4e9e1e61e3aa.ngrok.io -> http://localhost:4449
   

3. In Edge Impulse, go to Create impulse, then:

   1. Set the image width / height to 192 x 192 (this is the only resolution that works).
   2. click Add a processing block, click Add custom block and enter the URL from the previous step.
   3. Click Add a learning block, click Classification.
   4. Click Save impulse.

4. You now have pose estimation as a preprocessing block:

   

5. Train your model as usual 🚀

How to run this block (hosted in Edge Impulse)

Note: this flow is only available for enterprise customers.

1. Init the DSP block via:

   $ edge-impulse-blocks init
   
   # Pick "DSP block" for 'Choose a type of block'
   

2. Push the block via:

   $ edge-impulse-blocks push
   

3. In Edge Impulse open a project owned by your organization, go to Create impulse, then:

   1. Set the image width / height to 192 x 192 (this is the only resolution that works).
   2. click Add a processing block, select the block.
   3. Click Add a learning block, click Classification.
   4. Click Save impulse.

4. Follow the steps above!

Running on device

This block will run on Linux devices as-is. Just deploy as usual from the Studio.

Updating the model

Due to the size of the model and some unsupported ops it won't work on MCU in its current form. If you decide to train a smaller custom model, you'll need to replace model.tflite. You'll get feedback on the model through the 'On-device performance widget' in the Studio: