Achieving an overall 95%ish accuracy with the ongoing implementation is the goal. Once the MVP is robust enough, almost everyone with a free camera at 1080p will be able to use my MVP as an entirely free Steeldarts auto-scoring solution.
| Camera Angle | Camera Distance | Gaussian | Threshold | Erode | Accuracy | Success Ratio |
|---|---|---|---|---|---|---|
| 45° | 50 cm | 3 | 60 | 2 | 93.3% | 2:30 |
| 65° | 120 cm | 3 | 60 | 1 | 96.5% | 1:30 |
| 75° | 120 cm | 5 | 125 | 1 | 100% | 0:30 |
| 85° | 120 cm | 5 | 60 | 1 | 93.3% | 2:30 |
| 65° | 120 cm | 3 | 50 | 1 | 92.5% | 6:80 |
| 65° | 120 cm | 3 | 55 | 1 | 98.4% | 3:190 |
- at least 1080p static camera (smartphone camera might also work)
- camera positioned at the dartboard in an angle from 50° up to 80°
- Java 17
video comming soon
https://drive.google.com/drive/folders/1q4eLVK3SdFJBQlM-hmY64F3o52IK8G5j?usp=sharing
This approach is inspired by Jacob D. Delaney`s EE368/CS232 project proposal. Thanks for your interest.