Knowledge Base

Privacy Notice


All Resources

All Resources

How to make Impakt best recognize your repetitions

All Resources

How to make Impakt best recognize your repetitions

May 13, 2022

In this guide you’ll learn how Impakt recognizes your repetitions and what you can do to optimize your experience.

How Impakt detects your repetitions

Our technology analyzes the video it captures from your webcam to detect whether you are performing the exercise. A video is just a series of images (called frames) captured in rapid succession. Usually about 30 times per second. Our technology detects your body positions in each of these frames, which in sequence become a movement. It then compares that with what we have taught it the exercise should look like and counts a rep if the movement is similar enough.

Optimize your setup

Before minting, always do your research and use the resources available to ensure a safe and successful transaction. Here are a few tips and tools that can benefit your minting experience:


The cleaner the background, the better the recognition. If there is stuff directly behind you (like furniture) this can lead to worse detection, because in some cases it might get confused with your limbs.

✅ clean background

❌ clutter / clothes / furniture behind you


Generally, the more contrast between you and the background, the better the recognition. One exception is very dark clothes - when too much light is swallowed, it is difficult to detect where your joints are, which can diminish recognition accuracy.

✅ if your background is dark, wear light clothes

✅ if your background is light, wear darker clothes

❌ don’t wear very dark clothes


The better the lighting, the better the recognition. With bad lighting comes bad contrast and increased noise in videos. This makes it very difficult to detect anything. During the day, there is typically enough lighting just from daylight. When it gets darker, you may need to assist with turning on one or two lamps.

✅ turn on lamps if your video looks dark and noisy

Camera postioning

Recognition works best when the camera is positioned straight and level between knee- and hip-height in front of you. This way there is overall the least amount of angle distortion between your limbs from the perspective of the camera, and works well for both standing and floor exercises.

✅ position the camera so your whole body is in frame

✅ position the camera between knee- and hip-height

❌ don’t angle the camera (pointing up or down)


Your computer has to analyze many images per seconds for Impakt, thus generally the faster your computer, the better. That said, we also tested the app with computers from 2013 and it still works, but recognition of very fast movement was not as good as with higher end devices. List of things that can help with recognition .We will post detailed requirements, once we have more data.

💡 faster computer

💡 better graphics-card (dedicated GPU)

💡 higher quality camera

💡 wide-angle camera

Further ideas to improve your experience

These do not necessarily help with recognition, but can improve your overall experience.

💡 get an external wide-angle lens for your camera, so you don’t have to stand so far back for the whole body to be in frame

💡 use your phone as a camera (e.g. using, since it might be better quality than your built in webcam

💡 connect your laptop to a TV via HDMI, so you’ll better see what’s on screen

Optimize your movment

Movement in general : Our technology is specifically programmed to only count reps if the exercise is done with proper form and as intended - it won’t count if e.g. you do only half-reps. If you move too fast (particularly in bad lighting conditions and/or with a slow computer) it can happen that the video looks blurry or some crucial positions are not captured, which can lead to your reps not being recognized and counted.

✅ do each exercise with proper form (this also helps to avoid injuries)

✅ do the full range of motion

❌ don’t move too fast