Docker Wyze Bridge (redux)

WebRTC/RTSP/RTMP/HLS Bridge for Wyze Cam

Create a local WebRTC, RTSP, RTMP, or HLS/Low-Latency HLS stream for most of your Wyze cameras including the outdoor, doorbell, and 2K cams.

No modifications, third-party, or special firmware required.

It just works!

Streams direct from camera without additional bandwidth or subscriptions.

Please consider ⭐️ starring or ☕️ sponsoring this project if you found it useful, or use the affiliate link when shopping on amazon!

Important

As of May 2024, you will need an API Key and API ID from: Wyze Support Article.

[!WARNING] Please double check your router/firewall and do NOT forward ports or enable DMZ access to the bridge unless you know what you are doing!

See the supported cameras section for additional information.

Quick Start

Install docker and run:

docker run -p 8554:8554 -p 8888:8888 -p 5050:5000 -e WB_AUTH=false idisposablegithub365/wyze-bridge

You can then use the web interface at http://localhost:5050 where localhost is the hostname or ip of the machine running the bridge.

See basic usage for additional information or visit the wiki page for additional information on using the bridge as a Home Assistant Add-on.

What’s Changed in v3.12.3

Cleaned up the threading logic around startup/shutdown to reduce CPU and memory leaks

• Added validation checks for the API Key ID and API Key to help prevent issues logging in Fixes #47
• Cleaned up the thread and process tracking to ensure that we release threads when they’re done
• Only allow one running purge thread per camera Fixes #40
• Added timeouts to all the thread .join()s to ensure we don’t hang waiting for threads to die off
• Increased the buffer site for the pipe reads to reduce CPU load
• Consistently swallow ValueError, AttributeError, RuntimeError, and FileNotFound errors so sub-processes and threads terminate correctly

Note: v3.12.2 was everything above, but missing the change notes, oops.

What’s Changed in v3.12.1

Cleaned MQTT logic and pull in some others’ changes

New features

Automatic Sunrise/Sunset snapshots drive ny the LONGITUDE and LATITUDE configuration variables. FORCE_IOTC_DETAIL if true will force detailed debugging for the IOTC subsystem which can be used to decode a camera’s protocol messages.

• MQTT cleanup with more logging and move configuration to config.py

• Read the TUTK device_config.json once not every interaction

• Gathering up other changes

• Add GW_DBD Doorbell Duo to list not yet validated from @Angel-Vazquez change

• Add SNAPSHOT_CAMERAS and sunset/sunrise snapshots from @ruddell see

• Picking up the relevant changes from p2pcam

• Cleanup of config circular dependencies

• Fix run if syntax old habits die hard

What’s Changed in v3.12.0

Cleaned up the startup logic to ensure things start quickly and moved configurations around so everything is overrideable

• Moved API-driven snapshots into the Stream_Manager.py so they don’t delay startup.
• Clean up the /var/log directory in Dockerfile build image
• Better to supply empty, not samples in config
• Switch to run for diagnostic dump
• Deprecate the config.yml setting of environment variables so we can set the IMG_DIR and RECORD_PATH reliably even if not running inside Home Assistant as an add-on
• Remove the MTX_* variables from the .env as we want them to be settable in the options.
• Explicitly default the MTX settings that used to be forced (MTX_READTIMEOUT=30s, MTX_HLSVARIANT=mpegts, MTX_WRITEQUEUESIZE=2048) to ensure backward compatibility
• Updated the defaults for the IMG_DIR and RECORD_PATH to match what the .env would have previously set, but now it’s overridable.
• Reduced log spam by making some .info( calls .debug(
• Don’t emit ffmpeg non-error messages (let the Popen eat them).

What’s Changed in v3.11.1

Turns out you cannot have a completely optional section in a config.yml

Use something like

CAM_OPTIONS:
- CAM_NAME: fake-camera-name
RECORD: false

What’s Changed in v3.11.0

Cleanup of authorization logic and adding background activity

• RECORD is working again!
• Marked CAM_OPTIONS and MEDIAMTX as optional in the config
• Cleanup the snapshot pruning to ignore files going missing and use only prune each camera’s path, not the entire image directory
• Added background pruning of snapshots to speed startup
• Fix forced DEBUG log level
• Fix LOW_LATENCY should be LLHLS
• Extend session connection timeout to 60 seconds
• Fixed FPS calculation
• Split out WyzeStreamOptions
• Split out StreamManager
• Make Stream know type of camera and options.
• Make WyzeStream be a Stream
• Reduced default logging level for ffmpeg
• Tons of logging cleanup
• Cleaned up warnings

What’s Changed in v3.10.14

• Made MQTT config value optional Fixes #39
• Fix MQTT parameters minimum value for bitrate and fps
• Don’t emit MQTT state messages unless the state has changed
• Fix warning in BOA check
• Add missing OFFLINE_TIME and DOMAIN options
• Added documentation of the option defaults.

What’s Changed in v3.10.13

• Fix schema for MQTT discovery messages

What’s Changed in v3.10.12

• Filled out the missing translations and docs
• Fix busted device configuration table (the device.json file had some XML in it)
• Fix bad handling of WYZEDB3 message protocol support
• Clean up wyzecam to adopt original upstream MAIN fixes
• Sync with upstream wysecam DEV branch
• Fix sleep when no data is ready (was waiting 0 seconds instead of the intended 1/80th of a second)
• Yield frame_info in the receive data iterations.
• Decode both FrameInfoStruct OR FrameInfo3Struct and whine if something is wrong.
• Don’t discard messages without an expected_response_code as we know what it’s supposed to be anyway.
• Proper type, use literals, better logging, eliminate dead code
• Remove exit logging
• Revert type of resend and remove redundant SDK license set
• Add resp so debug structure are visible

What’s Changed in v3.10.11

• Fix errors in startup avClientStartEx doesn’t return a tuple
• Fix path construction for MediaMTX
• Fix python lint warnings and a whole lot of logging and type cleanup
• Move environment stuff to config and config reading out of mtx_server
• Don’t complain about directories existing in migration
• Attempt to ensure directories for recordings
• Capture stdout of MediaMTX and openssl for logging
• Capture the traceback before context corrupted
• Switched to better logging syntax
• Let’s declare the ports we EXPOSE in the Dockerfile(s)

What’s Changed in v3.10.10

• Add camera IP to MQTT message
• Adjust the recording path construction more
• Added STUN_SERVER support
• Switched all home assistant configs to host_network
• Don’t default WB_AUTH nor MQTT on and don’t force the MQTT_TOPIC
• Restored the fps to the K10056SetResolvingBit message
• Lots more logging to help track down the recording issue
• Cleanup a bunch of Python warnings
• Bump MediaMTX to 1.12.3
• Bump Wyze app version to 3.5.5.8
• Don’t force MediaMTX logging level to info
• Better tagging for Docker images
• Unified more the normal/hardware/multiarch docker build files
• Add devcontainer.json and tasks.json for VSCode

What’s Changed in v3.10.9

• Revert tutk_protocol change in K10056SetResolvingBit

What’s Changed in v3.10.8

• Removed forced leading "/" from RECORD_PATH

• Removed the IP restrictions from the MediaMTX "publisher" role

• Sync up with Elliot Kroo’s wyzecam library

• add HL_WCO2 camera support

• K10020CheckCameraParams support

• Fix authentication_type’s type

• Add fps to the K10056SetResolvingBit message

• Fix time setting to always advance one second (for lag) in K10092SetCameraTime

• Send/recv time and blank (PST flag?) to K11006GetCurCruisePoint/K11010GetCruisePoints/K11012SetCruisePoints

• Changed the MediaMTX config builder to emit correct config for recording

• Cleanup the warnings in the app code and added mtx_event pipe receipt logging

• Updated Wyze iOS app version to 3.5.0.8 (for user agent)

• Use SIGTERM for more graceful shutdown

• More startup logging for the MTX configuration of RECORD_PATH

• Sync up all the ports listed in MediaMTX with the ports exposed in the docker-compose files

What’s Changed in v3.10.7

• Reverted defaulting of RECORD_PATH option specifying {cam_name} instead of %path (need to fix that another way)
• Changed the MediaMTX config builder to emit correct config for recording.

What’s Changed in v3.10.6

• Changed the documentation and defaults for the RECORD_PATH option to specify {cam_name} instead of %path to eliminate recording errors Reverted in v3.10.7
• Add exception handling to ffmpeg pruning logic to prevent snapshot prunes from killing each other
• Now gathers the list of parents that might be pruned and does that after purging the files
• Fixed python lint message in get_livestream_cmd

What’s Changed in v3.10.5

• Fix regression for snapshot pruning

What’s Changed in v3.10.4

• Catch exceptions when pruning snapshots so we don’t stop grabbing them if something breaks a prune.
• Allow the ffmpeg error messages to reach the normal runtime
• Bump to MediaMTX 1.12.2 to fix regression on RaspberryPIs

What’s Changed in v3.10.3

• Bump MediaMTX to 1.12.1

What’s Changed in v3.10.2

• Added code to protect against the aggressive syntax check in MediaMTX 1.12.0 which complains about the recordPath missing required elements even when recording is not enabled (it really shouldn’t validate that setting unless one or more paths request recording...and didn’t through 1.11.3). For reference, the pattern is computed from our RECORD_PATH and RECORD_FILE_NAME settings and the combination of them must contain the strftime format specifiers of either a "%s" or all of of "%Y", "%m", "%d", "%H", "%M", "%S" (case-sensitive). If the value is not compliant, to keep MediaMTX from erroring out, we append "_%s" whatever was specified and emit a warning.
• Changed the default RECORD_PATH to "record/%path/%Y/%m/%d/" v3.10.7 "%path/{cam_name}/%Y/%m/%d"
• Changed the default RECORD_FILE_NAME to "%Y-%m-%d-%H-%M-%S"

What’s Changed in v3.10.1

• Add TOTP_KEY and MQTT_DTOPIC to config.yml schema to avoid logged warning noise
• Add MQTT_DTOPIC to config.yml options to ensure a usable default
• Add video: true to all the config.yml variants to ensure hardware encoding can use video card
• Upgrade to python:3.13-slim-bookworm for docker base image
• Cleaned up Dockerfile scripts for testing and multiarch
• Safer docker build by testing the tarballs downloaded for MediaMTX or FFMpeg

What’s Changed in v3.10.0

• Attempt upgrade of MediaMTX to 1.12.0 (again)
• Fixed schema of RECORD_LENGTH config option (it needs an s or h suffix, so must be string)
• Added RECORD_KEEP to the config.yml so it can be actually be configured in the add-on

What’s Changed in v3.0.7

• Better logging of exceptions and pass the MediaMTX messages through to main logs
• Correct building of permissions for MediaMTX
• Documented all the possible points in the docker-compose files.

What’s Changed in v3.0.6

• Revert MediaMTX to 1.11.3 because 1.12 doesn’t work here.

What’s Changed in v3.0.5 DELETED

• Fix MediaMTX to pass a user name since 1.12.0 now requires one

What’s Changed in v3.0.4 DELETED

• Chore: Bump MediaMTX to 1.12.0

What’s Changed in v3.0.3

Rehoming this to ensure it lives on since PR merges have stalled in the original (and most excellent) @mrlt8 repo, I am surfacing a new release with the PRs I know work. Note The badges on the GitHub repo may be broken and the donation links still go to @mrlt8 (as they should!)

• Chore: Bump Flask to 3.1.*
• Chore: Bump Pydantic to 2.11.*
• Chore: Bump Python-dotenv to 1.1.*
• Chore: Bump MediaMTX to 1.11.3
• FIX: Add host_network: true for use in Home Assistant by @jdeath to allow communications in Docker
• FIX: Hardware accelerated rotation by @giorgi1324
• Enhancement: Add more details to the cams.m3u8 endpoint by @IDisposable
• FIX: Fix mixed case when URI_MAC=true by @unlifelike
• Update: Update Homebridge-Camera-FFMpeg documentation link by @donavanbecker
• FIX: Add formatting of {cam_name} and {img} to webhooks.py by @traviswparker which was lost
• Chore: Adjust everything for move to my GitHub repo and Docker Hub account

What’s Changed in v2.10.3

• FIX: Increased MTX_WRITEQUEUESIZE to prevent issues with higher bitrates.

• FIX: Restart RTMP livestream on fail (#1333)

• FIX: Restore user data on bridge restart (#1334)

• NEW: SNAPSHOT_KEEP Option to delete old snapshots when saving snapshots with a timelapse-like custom format with SNAPSHOT_FORMAT. (#1330)

• Example for 3 min: SNAPSHOT_KEEP=180, SNAPSHOT_KEEP=180s, SNAPSHOT_KEEP=3m

• Example for 3 days: SNAPSHOT_KEEP=72h, SNAPSHOT_KEEP=3d

• Example for 3 weeks: SNAPSHOT_KEEP=21d, SNAPSHOT_KEEP=3w

• NEW: RESTREAMIO option for livestreaming via restream.io. (#1333)

• Example RESTREAMIO_FRONT_DOOR=re_My_Custom_Key123

What’s Changed in v2.10.2

• FIX: day/night FPS slowdown for V4 cameras (#1287) Thanks @cdoolin and @Answer-1!
• NEW: Update battery level in WebUI

What’s Changed in v2.10.0/v2.10.1

FIXED: Could not disable WB_AUTH if WB_API is set. (#1304)

WebUI Authentication

Simplify default credentials for the WebUI:

• This will not affect users who are setting their own WB_PASSWORD and WB_API.

• Default WB_PASSWORD will now be derived from the username part of the Wyze email address instead of using a randomly generated password.

• Example: For the email address john123@doe.com, the WB_PASSWORD will be john123.

• Default WB_API will be based on the wyze account for persistance.

Stream Authentication

NEW: STREAM_AUTH option to specify multiple users and paths:

Username and password should be separated by a :

An additional : can be used to specify the allowed IP address for the user.

• This does NOT work with docker desktop
• Specify multiple IPs using a comma

Use the @ to specify paths accessible to the user.

• Paths are optional for each user.
• Multiple paths can be specified by using a comma. If none are provided, the user will have access to all paths/streams

Multiple users can be specified by using | as a separator

EXAMPLE:

STREAM_AUTH=user:pass@cam-1,other-cam|second-user:password@just-one-cam|user3:pass

user:pass has access to cam-1 and other-cam

second-user:password has access to just-one-cam

user3:pass has access to all paths/cameras

See Wiki for more information and examples.

Recording via MediaMTX

Recoding streams has been updated to use MediaMTX with the option to delete older clips.

Use RECORD_ALL or RECORD_CAM_NAME to enable recording.

• RECORD_FILE_NAME Available variables are %path, {CAM_NAME}, {cam_name}, %Y %m %d %H %M %S %f %s (time in strftime format).
• RECORD_PATH Available variables are %path, {CAM_NAME}, {cam_name}, %Y %m %d %H %M %S %f %s (time in strftime format).
• RECORD_LENGTH Length of each clip. Use s for seconds , h for hours. Defaults to 60s
• RECORD_KEEP Delete older clips. Use s for seconds , h for hours. Set to 0s to disable automatic deletion. Defaults to 0s

Note that as of release v3.10.0, which uses mediaMTX 1.12.0, requires the combination of your RECORD_FILE_NAME and RECORD_PATH settings specifying recording’s file complete path MUST reference ALL of %Y %m %d %H %M %S tokens, for example: RECORD_FILE_NAME = "%H%M%S%" and RECORD_PATH_NAME = "{path}/{cam_name}/%Y/%m/%d%" would be valid. You can also just ensure the combined path includes %s (which is the unix epoch value as an integer).

View previous changes

FAQ

• How does this work?

• It uses the same SDK as the app to communicate directly with the cameras. See kroo/wyzecam for details.

• Does it use internet bandwidth when streaming?

• Not in most cases. The bridge will attempt to stream locally if possible but will fallback to streaming over the internet if you’re trying to stream from a different location or from a shared camera. See the wiki for more details.

• Can this work offline/can I block all wyze services?

• No. Streaming should continue to work without an active internet connection, but will probably stop working after some time as the cameras were not designed to be used without the cloud. Some camera commands also depend on the cloud and may not function without an active connection. See wz_mini_hacks for firmware level modification to run the camera offline.

• Why aren’t all wyze cams supported yet (OG/Doorbell Pro)?

• These cameras are using a different SDK and will require a different method to connect and stream. See the awesome cryze project by @carTloyal123.

Compatibility

Should work on most x64 systems as well as on most modern arm-based systems like the Raspberry Pi 3/4/5 or Apple Silicon M1/M2/M3.

The container can be run on its own, in Portainer, Unraid, as a Home Assistant Add-on, locally or remotely in the cloud.

Ubiquiti Unifi

Note

Some network adjustments may be needed - see this discussion for more information.

Supported Cameras

Important

Some newer camera firmware versions may cause issues with remote access via P2P. Local "LAN" access seems unaffected at this time. A temporary solution is to use a VPN. See the OpenVPN example.

Camera	Model	Tutk Support	Latest FW
Wyze Cam v1 [HD only]	WYZEC1	✅	3.9.4.x
Wyze Cam V2	WYZEC1-JZ	✅	4.9.9.x
Wyze Cam V3	WYZE_CAKP2JFUS	✅	4.36.11.x
Wyze Cam V4 [2K]	HL_CAM4	✅	4.52.3.x
Wyze Cam Floodlight	WYZE_CAKP2JFUS	✅	4.36.11.x
Wyze Cam Floodlight V2 [2k]	HL_CFL2	✅	4.53.2.x
Wyze Cam V3 Pro [2K]	HL_CAM3P	✅	4.58.11.x
Wyze Cam Pan	WYZECP1_JEF	✅	4.10.9.x
Wyze Cam Pan v2	HL_PAN2	✅	4.49.11.x
Wyze Cam Pan v3	HL_PAN3	✅	4.50.4.x
Wyze Cam Pan Pro [2K]	HL_PANP	✅	-
Wyze Cam Outdoor	WVOD1	✅	4.17.4.x
Wyze Cam Outdoor v2	HL_WCO2	✅	4.48.4.x
Wyze Cam Doorbell	WYZEDB3	✅	4.25.1.x
Wyze Cam Doorbell v2 [2K]	HL_DB2	✅	4.51.1.x
Wyze Cam Doorbell Pro 2	AN_RDB1	❓	-
Wyze Battery Cam Pro	AN_RSCW	⚠️	-
Wyze Cam Flood Light Pro [2K]	LD_CFP	⚠️	-
Wyze Cam Doorbell Pro	GW_BE1	⚠️	-
Wyze Cam OG	GW_GC1	⚠️	-
Wyze Cam OG Telephoto 3x	GW_GC2	⚠️	-

Basic Usage

docker-compose (recommended)

This is similar to the docker run command, but will save all your options in a yaml file.

1. Install Docker Compose.
2. Use the sample as a guide to create a docker-compose.yml file with your wyze credentials.
3. Run docker-compose up.

Once you’re happy with your config you can use docker-compose up -d to run it in detached mode.

Caution

If your credentials contain a $ character, you need to escape it with another $ sign (e.g., pa$$word > pa$$$$word) or leave your credentials blank and use the webUI to login.

[!NOTE] You may need to update the WebUI links if you’re changing the ports or using a reverse proxy.

Updating your container

To update your container, cd into the directory where your docker-compose.yml is located and run:

docker-compose pull # Pull new image
docker-compose up -d # Restart container in detached mode
docker image prune # Remove old images

🏠 Home Assistant

Visit the wiki page for additional information on Home Assistant.

Additional Info

• Camera Commands (MQTT/REST API)
• Two-Factor Authentication (2FA/MFA)
• ARM/Apple Silicon/Raspberry Pi
• Network Connection Modes
• Portainer
• Unraid
• Home Assistant
• Homebridge Camera FFmpeg
• HomeKit Secure Video
• WebUI API

Web-UI

The bridge features a basic Web-UI which can display a preview of all your cameras as well as direct links to all the video streams.

The web-ui can be accessed on the default port 5000:

http://localhost:5000/

See also:

• WebUI page
• WebUI API page

WebRTC

WebRTC should work automatically in Home Assistant mode, however, some additional configuration is required to get WebRTC working in the standard docker mode.

• WebRTC requires two additional ports to be configured in docker:

ports:
- 8889:8889 #WebRTC
- 8189:8189/udp # WebRTC/ICE

• In addition, the WB_IP env needs to be set with the IP address of the server running the bridge.

environment:
- WB_IP=192.168.1.116

• See documentation for additional information/options.

Advanced Options

All environment variables are optional.

• Audio
• Bitrate and Resolution
• Camera Substreams
• MQTT Configuration
• Filtering Cameras
• Doorbell/Camera Rotation
• Custom FFmpeg Commands
• Interval Snapshots
• Stream Recording and Livestreaming
• rtsp-simple-server/MediaMTX Config
• Offline/IFTTT Webhook
• Proxy Stream from RTSP Firmware
• BOA HTTP Server/Motion Alerts
• Debugging Options

Other Wyze Projects

Honorable Mentions:

• @noelhibbard’s script - Original script that the bridge is bassd on.
• kroo/wyzecam - Original library that the bridge is based on.

Video Streaming:

• gtxaspec/wz_mini_hacks - Firmware level modification for Ingenic based cameras with an RTSP server and self-hosted mode to use the cameras without the wyze services.
• thingino - Advanced custom firmware for some Ingenic-based wyze cameras.
• carTloyal123/cryze - Stream video from wyze cameras (Gwell cameras) that use the Iotvideo SDK from Tencent Cloud.
• xerootg/cryze_v2 - Stream video from wyze cameras (Gwell cameras) that use the Iotvideo SDK from Tencent Cloud.
• mnakada/atomcam_tools - Video streaming for Wyze v3.

General Wyze:

• shauntarves/wyze-sdk - python library to interact with wyze devices over the cloud.