Comparing to competitors
Accuracy comparison video between conventional depth sensors and VisionPose.
Real-time accuracy demo of three skeletal detection systems
We compared the real-time detection accuracy of three skeletal detection systems, including VisionPose.
*To accurately compare the performance, we got rid of any delay due to difference of processing power.
Accuracy comparison video between depth sensor camera and VisionPose on a sunny day
Left: Competing depth sensor camera / Right: VisionPose
This video compares the real-time accuracy of VisionPose to an existing depth sensor camera, which has always faced issues with footage taken from the outside.
Accuracy comparison video between depth sensor camera and VisionPose in real-time
Left: Competing depth sensor camera / Right: VisionPose
We compared conventional depth sensor cameras and their ability to deliver real-time posture inference with VisionPose. Conventional depth sensor cameras often have trouble with movement where limbs are crossing, or the detection of legs.
Key point comparison table
| VisionPose | CompetitorA | CompetitorB | |
|---|---|---|---|
| Compatible Platforms | |||
| Compatible Platforms | ■Standard Windows 10(*) Ubuntu 18.04(*) *Able to be used in the Cloud ■Single3D Windows Unity iOS/iPadOS | Windows 8/10 Ubuntu 14/16 Mac OSX NVIDIA TX2 | Not mentioned |
| Usage Costs | |||
| Usage Costs | VisionPose Standard: $4,950 VisionPose Single3D: $4,950 | Commercial use: $25,000 a year Non-commercial use: Free *Cannot be used for sports. | GNU General Public License v2.0 |
| Delivered as | |||
| Delivered as | DownLoad | Source code | Source code |
| 3D Inference | |||
| 3D Inference | Yes | Partially* | No |
| Key Points | |||
| Key Points | 30 | 15,18,25 | Not mentioned |
| Single Camera FPS** | |||
| Single Camera FPS** | Standard: ~60.0 Single3D Unity: ~60.0 Single3D iOS/iPad: ~15.0 | ~9.75 | Not mentioned |
| Dual Camera FPS** | |||
| Dual Camera FPS** | Standard：~59.6 | N/A | N/A |
| Detectable Amount of People | |||
| Detectable Amount of People | Up to 16 people | Not mentioned | Not mentioned |
| Dual Camera Placement | |||
| Dual Camera Placement | You can put the cameras in a parallel configuration, or put them at an angle. Parallel setup When putting the webcams in a parallel setup, keep the following in mind. Lenses: Roughly 60 centimeters apart Cameras should be around 2.5 to 3 meters removed from subject. Angled setup Two webcams surround the subject by being placed at an angle to one another. Keep in mind that this angle can at most be 90 degrees (for example: one camera in the front and one on the side). | Not mentioned | Not mentioned |
- *Offered as a separate module
- **Specs of PC or phone can alter performance.