Continuous gestures

This is a prebuilt dataset for a gesture recognition system based on continuous motion, for the Continuous motion recognition tutorial. It contains 15 minutes of data sampled from a MEMS accelerometer at 62.5Hz over the following four classes:

  • Idle - board sits idly on your desk. There might be some movement detected, e.g. from typing while the board is present.
  • Snake - board moves over the desk as a snake.
  • Updown - board moves up and down in a continuous motion.
  • Wave - board moves left and right like you're waving to someone.

Demonstration movements

IdleIdle

Idle

SnakeSnake

Snake

UpdownUpdown

Updown

WaveWave

Wave

Importing this dataset

You can import this dataset to your Edge Impulse project using the Edge Impulse CLI Uploader. If you haven't done so, follow the Installation instructions.

Then:

  1. Download the gestures dataset.
  2. Unzip the file in a location of your choice.
  3. Open a terminal or command prompt, and navigate to the place where you extracted the file.
  4. Run:
$ edge-impulse-uploader --clean
$ edge-impulse-uploader --category training training/*.cbor
$ edge-impulse-uploader --category testing testing/*.cbor

You will be prompted for your username, password, and the project where you want to add the dataset.

Capturing all variations

Naturally, this dataset does not capture all variations of the motions. You might be doing these motions different than us. You can easily make the model more resilient by capturing more data from the Data acquisition page.


Did this page help you?