These sample .nml files are for training a classification model using video data in NeoPulse® AI Studio.

Data for this example is from the Human Actions Dataset. The dataset features 25 subjects performing 6 actions in 4 different scenarios.

In order to use the build_csv.py script, you will need pip3. If you don't already have it, you can install it using the command apt install python3-pip After that, you will need to install natsort. To do that, you will need to run: pip3 install natsort

To run this example, first you will need to download and pre-process the raw data for the task using the included build_csv.py script:

$ python3 build_csv.py

The script will not execute using Python 2.7.

If the script fails, make sure that you have installed all the package dependencies of this script which are listed at the top of the script: zipfile, shutil, pathlib, requests, random, and natsort. Missing packages can be installed using pip:

$ pip3 install <package_name>

Once you've downloaded and pre-processed the data, you can start training using any of the NML scripts provided. To begin training:

$ neopulse train -p <project_name> -path /absolute/path/to/video_class_auto.nml

In NML file, relative path of .csv file is declared in the line:

bind = "training_data.csv" ;

Tutorial videos are available in the Tutorials section of the AIDynamics® Developer Portal

For more information on using the VideoDataGenerator visit the [Data section] of the NeoPulse® AI Studio Documentation(https://docs.neopulse.ai/NML-source/#data)

video_class_auto.nml: This classification model features full use of the auto keyword to automatically generate the entire architecture.

video_class.nml: This classification model features a more advanced architecture without use of the auto keyword.

Tutorial materials are published under the MIT license. See license for commercial, academic, and personal use.

You are welcome to modify these tutorial files. If citing please link to this repository.