# Setting Up the Capture System
This section will focus on setting up the capture system in order to begin data collection. In a typical project this requires some work. It's good to think about what tools you have to work with, what you are actually trying to capture, how will you capture it, and how the final model will actually work. In this case we will answer such questions as follows:
- We want to detect gestures using a radar
- We will define gestures that can be done continuously so that we can collect data files as long recordings of the same gesture continuously: Vertical Finger Rotation, Push and Wiggle
- The labels will then be the entire video of the continuous gesture
- We will use the Acconeer radar set up to stream data through a serial link
- The serial data is then read through the capture server and combined with a video from the capture app to allow for easy labelling
- The generated model is then only inspected in the studio and not tested on the hardware (if that was also a part of it, one would need to solve how to integrate the AI model with the sensor and run it on a practical device)
For starting a new project one should always dig into this as deep as possible to identify possible shortcomings or problems that would need to be solved in advance.
The first step is to setup the hardware. You should already have the:
Acconeer XM122 module
Acconeer XB122 breakout board
Follow the following guide (opens new window) by Acconeer to flash the device with the correct firmware. This is required in order to progress. You could also follow this video (opens new window) made by acconeer.
You'll need to first download the Capture Server from here (opens new window) and install it according to the provided instructions (opens new window). It is to be noted that you should pull the whole CaptureServer folder so you can run the Acconeer example.
Once the Imagimob Capture Server has been properly installed. We will need to ensure that the hardware is set up according to guide (opens new window). This example has pre-built functionality that's compatible with the XM122 IoT Module.