ActionAI is a python library for training machine learning models to classify human action. It is a generalization of our yoga smart personal trainer, which is included in this repo as an example.

These instructions will show how to prepare your image data, train a model, and deploy the model to classify human action from image samples. See deployment for notes on how to deploy the project on a live stream.

Docker installation is recommended:

The included Dockerfile builds for Jetson devices running Jetpack 4.6.1. To build, cd into the docker/ directory and run:

docker build -f jetson-deployment.dockerfile -t actionai:j4.6.1 .

You can also pull a prebuilt image hosted on Docker Hub.

docker pull smellslikeml/actionai:j4.6.1

docker run -itd --rm 
         --net=host 
         --privileged 
         --env=DISPLAY 
         --runtime=nvidia  # for GPU
         --env=QT_X11_NO_MITSHM=1   # for visualization
         -v /tmp/.X11-unix:/tmp/.X11-unix 
	 --device /dev/input/js0      # for PS3 controller
	 -v /run/udev/data:/run/udev/data
         -v /dev/bus/usb:/dev/bus/usb  # for depthai camera
         --device-cgroup-rule='c *:* rmw' 
         -v /path/to/ActionAI:/app/ 
         smellslikeml/actionai:j4.6.1-latest /bin/bash

Alternatively, use a virtual environment to avoid any conflicts with your system's global configuration. You can install the required dependencies via pip:

We use the trt_pose repo to extract pose estimations. Please look to this repo to install the required dependencies. You will also need to download these zipped model assets and unzip the package into the models/ directory.

# Assuming your python path points to python 3.x 
$ pip install -r requirements.txt

We've provided a sample inference script, inference.py, that will read input from a webcam, mp4, or rstp stream, run inference on each frame, and print inference results.

If you are running on a Jetson Nano, you can try running the iva.py script, which will perform multi-person tracking and activity recognition like the demo gif above Getting Started. Make sure you have followed the Jetson Nano installation instructions above and simply run:

$ python iva.py 0

# or if you have a video file

$ python iva.py /path/to/file.mp4

If specified, this script will write a labeled video as out.mp4. This demo uses a sample model called lstm_spin_squat.h5 to classify spinning vs. squatting. Change the model and motion dictionary under the RUNSECONDARY flag to run your own classifier.

We've also included a script under the experimental folder, online_finetune.py, that supports labelling samples via a PS3 Controller on a Jetson Nano and training in real-time from a webcam stream. This will require these extra dependencies:

• Pygame
• PS3 Controller

To test it, run:

# Using a webcam
$ python experimental/online_finetune.py /dev/video0  

# Using a video asset
$ python experimental/online_finetune.py /path/to/file.mp4  

This script will also write labelled data into a csv file stored in data/ directory and produce a video asset out.mp4.

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

This project is licensed under the GNU General Public License v3.0 - see the LICENSE.md file for details

• Hackster post
• YogAI article
• Convolutional Pose Machine
• Pose estimation for mobile
• Pose estimation tensorflow implementation