dearsky/wifi-densepose
1
0
Fork 0
Code Issues 20 Pull Requests 5 Actions Packages Projects Releases 1 Wiki Activity

Is this a real and usable project？ #79

New Issue
Open
opened 2026-03-02 11:43:56 +08:00 by AfricanCh · 5 comments
AfricanCh commented 2026-03-02 11:43:56 +08:00 (Migrated from github.com)
Copy Link

I haven't tried it yet, which requires some cost. I have seen many issues with reported issues being closed as' unplanned ', which makes me cautious. Please forgive me if I have offended you

I haven't tried it yet, which requires some cost. I have seen many issues with reported issues being closed as' unplanned ', which makes me cautious. Please forgive me if I have offended you
p20061 commented 2026-03-02 12:09:33 +08:00 (Migrated from github.com)
Copy Link

我也感觉很奇幻，那这么来说蓝牙应该也可以，只是长的电磁波

我也感觉很奇幻，那这么来说蓝牙应该也可以，只是长的电磁波
oki-zon3 commented 2026-03-02 12:36:38 +08:00 (Migrated from github.com)
Copy Link

Philips, with its WiZ smart Wi-Fi products, uses a technology similar to Spacesense, which is Wi-Fi-based motion detection. It works by having at least two products (e.g., smart Wi-Fi lamps) communicate continuously in the same room. When an object moves in the room, a small change occurs in the communication process between the devices. This change is then analyzed by the device firmware as motion detection.

Philips, with its WiZ smart Wi-Fi products, uses a technology similar to Spacesense, which is Wi-Fi-based motion detection. It works by having at least two products (e.g., smart Wi-Fi lamps) communicate continuously in the same room. When an object moves in the room, a small change occurs in the communication process between the devices. This change is then analyzed by the device firmware as motion detection.
👎 1
36mimu36 commented 2026-03-02 18:08:15 +08:00 (Migrated from github.com)
Copy Link

Yes, it works — tested with real hardware

I deployed the Python sensing pipeline on real hardware and can confirm it detects human presence and motion.

Hardware

  • 2x ESP32-S3-DevKitC (about $10 each)
  • 1x Raspberry Pi (running Debian 13)
  • Standard home WiFi router

What works

  • Presence detection: Correctly identifies person in room (PRESENT_STILL, 80% confidence)
  • Motion detection: Motion Band power rises when person moves (0.33 measured)
  • Multi-node: Two ESP32 nodes streaming CSI simultaneously, cross-receiver agreement boosts confidence
  • Real-time UI: Web UI at http://<pi>:8080 shows live signal field, RSSI, classification

Setup steps (bare metal, no Docker)

  1. Flash ESP32-S3 with the firmware in firmware/esp32-csi-node/ using ESP-IDF v5.2
  2. Set WiFi SSID/password and target IP in sdkconfig.defaults
  3. On Pi: install Python deps (numpy, scipy, websockets, pydantic), run ws_server.py
  4. ESP32 auto-streams ADR-018 binary CSI frames over UDP to the Pi

Total cost: ~$30 (2x ESP32-S3) + a Pi you may already have.

Caveats

  • Breathing detection (0.1–0.5 Hz) does not work yet with the Python pipeline — needs per-subcarrier phase tracking
  • The default mean_amplitude signal is noisy; I switched to subcarrier spread (np.std(amplitudes)) for better results (see my comment on #51)
  • Full DensePose (body mesh) requires trained neural network models not yet available — current capability is presence + motion level classification

It's not magic — it's WiFi CSI (Channel State Information), the same principle behind Philips WiZ Spacesense mentioned above. This project provides an open-source implementation.

### Yes, it works — tested with real hardware I deployed the Python sensing pipeline on real hardware and can confirm it detects human presence and motion. #### Hardware - 2x ESP32-S3-DevKitC (about $10 each) - 1x Raspberry Pi (running Debian 13) - Standard home WiFi router #### What works - **Presence detection**: Correctly identifies person in room (PRESENT_STILL, 80% confidence) - **Motion detection**: Motion Band power rises when person moves (0.33 measured) - **Multi-node**: Two ESP32 nodes streaming CSI simultaneously, cross-receiver agreement boosts confidence - **Real-time UI**: Web UI at `http://<pi>:8080` shows live signal field, RSSI, classification #### Setup steps (bare metal, no Docker) 1. Flash ESP32-S3 with the firmware in `firmware/esp32-csi-node/` using ESP-IDF v5.2 2. Set WiFi SSID/password and target IP in `sdkconfig.defaults` 3. On Pi: install Python deps (`numpy`, `scipy`, `websockets`, `pydantic`), run `ws_server.py` 4. ESP32 auto-streams ADR-018 binary CSI frames over UDP to the Pi Total cost: ~$30 (2x ESP32-S3) + a Pi you may already have. #### Caveats - Breathing detection (0.1–0.5 Hz) does not work yet with the Python pipeline — needs per-subcarrier phase tracking - The default `mean_amplitude` signal is noisy; I switched to subcarrier spread (`np.std(amplitudes)`) for better results (see my comment on #51) - Full DensePose (body mesh) requires trained neural network models not yet available — current capability is presence + motion level classification It's not magic — it's WiFi CSI (Channel State Information), the same principle behind Philips WiZ Spacesense mentioned above. This project provides an open-source implementation.
👍 3 👎 6
ruvnet commented 2026-03-02 20:22:21 +08:00 (Migrated from github.com)
Copy Link

I'm no longer actively building the Python version. I should note almost everyone who said it doesn't work didn't actually use CSI compatible hardware.

Thanks for the confirmation.

I'm no longer actively building the Python version. I should note almost everyone who said it doesn't work didn't actually use CSI compatible hardware. Thanks for the confirmation.
👍 2
rezzorix commented 2026-03-02 20:25:09 +08:00 (Migrated from github.com)
Copy Link

almost everyone who said it doesn't work didn't actually use CSI compatible hardware.

May i suggest you make it more prominent in the initial description or right below the header in BOLD that certain hardware is required?

Bonus: maybe provide an example list of hardware that is working.

> almost everyone who said it doesn't work didn't actually use CSI compatible hardware. > May i suggest you make it more prominent in the initial description or right below the header in BOLD that certain hardware is required? Bonus: maybe provide an example list of hardware that is working.
👍 1
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
Something isn't working
 documentation
Improvements or additions to documentation
 duplicate
This issue or pull request already exists
 enhancement
New feature or request
 good first issue
Good for newcomers
 help wanted
Extra attention is needed
 invalid
This doesn't seem right
 question
Further information is requested
 security
Security-related issues and hardening
 wontfix
This will not be worked on
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
dearsky
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: dearsky/wifi-densepose#79
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?