tofcamis an optimized implementation of a time-of-flight depth camera library.- It computes depth and confidence (derived from amplitude) image from raw four-phase camera data.
- Raspberry Pi5
- Arducam ToF Camera
- For the initial setup, configure the environment according to the official Arducam documentation and confirm that the provided examples run correctly.
- For subsequent setups, the media graph can be configured using setup_rpi5.sh
- You have to execute below in your RPi.
$ git clone https://github.com/void-hoge/tofcam.git
$ cd tofcam
$ cmake -S . -B build
$ cmake --build build
- Values shown in parentheses indicate results when only depth is computed, without computing confidence.
- The frame rate indicates the number of raw frames processed per second, Four raw frames are combined to produce one depth image.
- CPU frequency fixed at 2.4GHz.
| Arducam ToF SDK | tofcam std::atan2 | tofcam approx_atan2 | tofcam neon | |
|---|---|---|---|---|
| CPU Load | 18.4% | 4.7% (4.3%) | 1.20% (1.11%) | 0.69% (0.57%) |
| Max Cvt Rate | (not supported) | 2600fps (2800fps) | 11500fps (12300fps) | 22900fps (28300fps) |
- CPU frequency fixed at 1.0GHz.
| Arducam ToF SDK | tofcam std::atan2 | tofcam approx_atan2 | tofcam neon | |
|---|---|---|---|---|
| CPU Load | 46.1% | 35.8% (31.1%) | 6.9% (6.1%) | 6.0% (4.6%) |
| Max Cvt Rate | (not supported) | 350fps (400fps) | 2300fps (2500fps) | 3000fps (4500fps) |
- Mugi Noda (void-hoge)
- MIT
