Always improving
What's new
We keep refining SkyView — here's a timeline of the features and improvements each release brings you.
- v0.9.60Latest
115 cloud storage overwrites the oldest recordings when nearly full · AI detection no longer takes minutes to start on hard drives · Detection uses less CPU
- Local and 115 cloud recordings are now managed separately. The Storage page is split into cards for this device, 115 cloud, retention rules and storage maintenance, and each storage location has its own "max space". Recordings on 115 can now have a "Max space for recordings"; beyond that, the oldest recordings are deleted.
- ⚠️ When 115 cloud storage is nearly full, the oldest recordings are now deleted automatically. Previously recordings on 115 were never removed, so once it filled up new recordings could no longer be uploaded. Now, when free space runs low, the oldest recordings are deleted to make room for new ones, the same way local storage loops. If 115 is mostly filled with your own other files, Yunkan does not delete recordings and only warns you. Recordings covered by events you chose to keep are never deleted.
- This device no longer fills up when "Delete the copy on this device after uploading to 115" is off. The local copy now counts toward this device's space limit and the oldest copies are removed first when it is full; the copy on 115 is not affected. Previously nothing ever removed these local copies, which could fill the drive.
- AI detection no longer takes several minutes to start after a restart on hard drives. Previously every startup checked all recordings and snapshots on disk one by one, which could keep a busy hard drive fully occupied for minutes while everything else waited. Startup no longer does this scan.
- Leftover recording cleanup is now done manually from the Storage page. Recording files that have no database record are no longer deleted automatically in the background; scan and confirm the cleanup under Storage maintenance. If an unusually large share of recordings has no record (for example after switching databases), you get an extra warning before anything is deleted. Recordings and snapshots left behind by a deleted camera are cleaned up automatically.
- AI detection uses less CPU. CPU usage drops noticeably when several cameras are detecting at once, and areas that keep moving but never contain anything (such as trees or a TV) are no longer checked on every frame.
- No more repeated "detection stopped" alerts while a camera is offline. The alert is no longer sent while the camera is disconnected, and detection resumes right away once the camera is back.
- More accurate message when marking a false positive. Only events such as people, vehicles, animals and packages go into training annotation; for falls, faces, crying and similar events the message no longer says the event will appear on the training annotation page.
- The one-line install script can show English output. Add
--lang enwhen installing; the default is still Chinese.
- v0.9.59Highlight
Hear the other side while talking on TP-Link cameras · Push-to-talk works on Dahua cameras · iOS app adapted for iPad and iPhone Duo
- Hear what is happening while you talk on TP-Link cameras. Previously you could not hear the camera side while holding the talk button, only after letting go. You can now hear both ways while talking, so a conversation feels natural. Upgrading the server is enough; no new phone app needed.
- Fixed silent push-to-talk on some Dahua cameras. The previous release fixed talk-down and voice announcements on these cameras, but holding the talk button on the phone or the web still produced no sound from the camera. Push-to-talk now plays normally too. Upgrading the server is enough; no new phone app needed.
- iOS app updated to 0.9.59 (available once App Store review completes), bringing together the iPhone and iPad improvements since 0.9.54. ⚠️ From this version the app requires iOS 18 or later; devices on iOS 16 or 17 will not receive this update, and the installed app keeps working as before.
- Adapted for the large screens of iPad and iPhone Duo: Live view and Playback now show the picture on the left and the controls on the right; landscape full screen is easier to use; the picture keeps playing when you fold, unfold or resize the window; multiple windows are supported, so you can open a single camera in its own window.
- Record the live view or playback you are watching right in the app. What you see is what gets recorded, saved to the Photos library, up to 30 minutes per recording.
- Find a person in Events using a photo (admin account and Pro license required), and use "Keep this event" in event details so the event and its recording are never removed by automatic cleanup.
- Lower latency when watching live; playback stops when the app goes to the background or the screen locks, and reconnects automatically when you come back.
- Fixed live view staying silent after a talk session ends; after switching the language inside the app, all text in the interface now switches with it.
- Android app updated to 0.9.59 to keep the version in step with the server. No functional changes.
- v0.9.58Highlight
Recordings no longer drop out for minutes at a time · Talk-down works on Dahua cameras · Activate a new license key right after it expires · Better picture and detection on Xiaomi cameras
- Recordings on some cameras no longer drop out for several minutes. On cameras that use G.711 audio (such as some TP-Link models), an unstable network used to leave gaps of several minutes in the recording even though the camera never went offline. Recording now picks up as soon as the camera is back, so the gap is only the few seconds it was actually disconnected.
- Fixed silent talk-down on some Dahua cameras. On some Dahua and Dahua-based cameras, push-to-talk, automation voice announcements and the talk-down preview showed as played but the camera made no sound. They now play normally.
- Activate a new license key right after your license expires. Previously, tapping "Activate now" in the classic interface after expiry took you back to the home page. You now stay on the activation page and can enter the new key. If your license is still valid and you want to renew early or switch to a new key, use "Change license key" under Settings → License, in both the classic and desktop interfaces.
- More accurate low-resolution detection on Xiaomi cameras. Some Xiaomi cameras that can provide a low-resolution stream were judged unable to, so AI detection had to use the main stream. The check is now more accurate: cameras previously judged unable are checked again once after upgrading, and on some models recording may pause for a few seconds during that check.
- Fixed a main stream stuck at 720p on certain Xiaomi cameras. On a few models the main picture was only 720p after adding. After upgrading it returns to the highest resolution the camera supports (such as 2K), with no need to add the camera again.
- Android gallery folders now follow the app language. Screenshots, event images and in-app recordings used to go into a folder named "云瞰" regardless of language. The folder name now follows the language set in the app: "云瞰" in Simplified Chinese, "雲瞰" in Traditional Chinese and "SkyView" in other languages. Requires the Android app 0.9.58.
- v0.9.57Highlight
Filter events by date · 115 cloud uploads no longer get stuck on one recording · Xiaomi cameras use fewer resources for detection and no longer show garbled frames · Faster detection on the Intel edition
- Filter the event center by date range. The event center on the web (classic and desktop) now has start and end dates, so you can look at a single day or a few days, or fill in just one end. It works together with the camera filter, search and finding a person from a photo, and scrolling down keeps loading every event in that period.
- 115 cloud uploads no longer get stuck on the same recording. Some recordings had in fact reached 115, but because the file details from the cloud drive never came back, they were retried over and over and kept showing as failed. Yunkan now recovers them automatically, and after upgrading these recordings sort themselves out. They stayed on local storage the whole time, so nothing was lost.
- Xiaomi cameras use a low-resolution stream for AI detection automatically. When a camera is added, Yunkan checks once whether it can also provide a low-resolution stream; if it can, AI detection uses that stream and needs less CPU, otherwise it keeps using the main stream. Xiaomi cameras you added earlier are checked once after upgrading; on some models recording may pause for a few seconds during that check, and each camera is only checked once.
- Fixed garbled frames in AI detection on some Xiaomi / Mi Home cameras. Some models switch resolution a few seconds after the stream starts, after which the picture used for AI detection was scrambled and detection was unreliable. Detection now follows the change in picture size.
- Faster detection on the Intel edition when the inference backend is set to Auto. With "Inference backend" set to "Auto (recommended)", detection used to end up on the slowest way of running; it now runs on the CPU in a faster way, making the same detection 1.2 to 4 times faster with identical results. To run inference on the integrated GPU, choose "Intel iGPU" under "Inference backend" as before. The resource self-check now shows where inference actually runs and warns you if you chose the GPU but it isn't being used; and if an acceleration method fails to start, detection switches to the CPU and keeps going instead of stopping for that camera.
- More reliable Intel GPU hardware decoding. Cameras whose own video stream is faulty are no longer mistaken for unstable GPU decoding and switched to software decoding; and if GPU decoding hits an error partway through on a particular camera, it continues on the GPU in a more compatible way.
- Talk-down and voice announcements fail less often. When speech synthesis occasionally fails because of a network hiccup, it now retries automatically a few times.
- v0.9.56Highlight
Add Xiaomi / Mi Home cameras directly · Hailo AI accelerator support · Find a person in events from a photo · Record what you are watching in the Android app · Lower live-view latency
- Add Xiaomi / Mi Home cameras directly. When adding a camera, choose "Xiaomi / Mi Home", sign in by scanning the QR code with the Mi Home app (or with your account and password), and pick the cameras on your account. Live view, recording, AI detection, pan/tilt and two-way audio all work, with no separate conversion tool to set up. Before adding, note that connecting to a camera needs an internet connection to verify with Xiaomi; each camera allows only one connection at a time, so watching live view in the Mi Home app may briefly interrupt the picture here; and we recommend giving the camera a fixed IP address on your router.
- Support for Hailo-8 / Hailo-8L AI accelerators. Plug in the accelerator and install its driver on the host with a single command; suitable features such as object detection and face detection then run on the accelerator automatically, so detection is faster and the load on the CPU and the chip's built-in NPU drops, while everything else runs as before. Settings → Detection shows where each feature runs, and if the accelerator can't be used yet, it shows the exact command to run on the host. Supported in the standard, Intel and RK3588 editions, but not the NVIDIA edition; it has been tested on RK3588 devices so far.
- Find a person in the event center from a photo. Upload a photo, pick a face in it, and the event center shows the events where that person appeared, with a count, and you can jump straight to playback. Available on the web (classic and desktop) and Android for administrators with a Pro licence; the photo is only used for that search and is not stored.
- Record what you are watching in the Android app. Both live view and playback can be recorded with one tap, exactly as shown on screen, and saved to the phone's gallery under "Movies/云瞰", up to 30 minutes per recording. Requires Android app 0.9.56.
- Lower live-view latency. Live view on the web and Android has noticeably less delay, lowest when muted; on some cameras (for example those using G.711 audio), turning the sound on no longer slows the picture down. The live-view dialog under Devices in the classic interface now has a mute button and starts muted.
- Android stops playback when the app goes to the background or the screen locks. Previously the app kept receiving video and playing sound in the background. Now live view disconnects and reconnects automatically when you come back, and recordings and event clips pause and resume where they left off.
- Android shows the actual reason when playback fails, instead of just an English "Source error": video not found, no permission, a network problem, or a format this phone can't decode.
- Jumping to the event center from the pet library and other pages now finds older events too. Previously only recent events could be located; older ones did nothing.
- On the labelling page, the "to label" count now goes down as you save, and the next batch loads automatically when one is finished. Previously the count only went down when you deleted an image, and finishing a batch of 100 showed "Nothing to label".
- v0.9.55Highlight
Talk-down audio works again · Oldest recordings are overwritten when the disk is nearly full, so recording never stops · Keep important events · The pet "To sort" queue no longer piles up
- Camera talk-down and spoken automation announcements work again. Since 13 October, talk-down previews and spoken announcements in automations were all failing, because the speech service changed how it accepts connections. Yunkan now supports the new method, and updating restores the feature. Two-way audio was not affected.
- ⚠️ When the disk is nearly full, the oldest recordings are deleted automatically so recording never stops. Previously, if the disk was filled by files outside Yunkan (downloads, other apps and so on), Yunkan deleted none of its recordings and recording eventually stopped. Now, when free space drops to roughly one hour of recording (2–3% of the disk), the oldest recordings are deleted to make room, the way a typical NVR does, and recording continues. When space is extremely tight, recordings not yet uploaded to cloud storage, and even recordings of events you chose to keep, can be deleted too. If Yunkan shares a disk with other apps, we recommend setting a total recording quota under Storage.
- Keep important events. Choose "Keep" in the event details ("Keep this event" on Android) and that event and its recording will not be removed by the day-based or quota-based automatic cleanup (except when the disk is about to fill up). Available on the web (classic and desktop) and Android; Android needs to be updated to 0.9.55.
- Keep event recordings longer than regular recordings. Storage settings gain "Keep event recordings for N days (optional)", so the footage around an event can be kept for longer on its own.
- See how long your recordings will last on the Storage page. It shows the oldest recording, roughly how much is written each day, how many days the current settings are expected to keep, and which setting is the limiting one. You can fill in a suggested total recording quota with one click (save it yourself afterwards), and you are reminded when other files take up much of the disk and no quota is set. When editing a camera you can also see its average bitrate, daily usage, space used so far and how many days it is expected to keep.
- The pet "To sort" queue no longer keeps growing. The system now keeps the queue in check by itself (at most 600 photos per camera, and candidates that already have enough photos stop collecting more). Candidates you delete will not be suggested again; the pet library shows "N animals you deleted won't be suggested again", and "Bring back suggestions" restores them. The queue is now a single line with a count and "View"; the dialog groups photos by camera with pages, and lets you delete selected photos, clear one camera or clear everything. After updating, devices with a large backlog will tidy it up once automatically.
- Merge a pet the system found into an existing pet. Choose "Merge into…" on the candidate card; the classic interface also lets you select several and merge them at once.
- Vehicle automations filtered by licence plate trigger again. Vehicle rules with licence plates filled in never fired before; this is fixed, and plates are now compared ignoring case, spaces and separators.
- Fixed the blank progress banner on the Training page that stopped updating while a model was being activated.
- Fixed Android Playback getting stuck when skipping back or forward 15 seconds across two recording segments. Skips now cross segment boundaries, and gaps without recordings are skipped in the direction you are going.
- v0.9.54Highlight
End-to-end encrypted phone notifications · Tell family from strangers even without a clear face · Simpler detection settings · Adjustable text size · Mac and ARM hosts supported
- Phone notifications are now end-to-end encrypted. Each notification is encrypted on your device separately for every phone, so only your phone can read it; the push services in between only ever see a placeholder line, and the attached snapshot can likewise only be fetched by your phone. The device itself no longer ships with any push credentials. The new Android and iOS apps turn this on automatically, with nothing to configure.
- ⚠️ Please update the mobile app to 0.9.54 as well. Phones still on an older app will by default only receive "You have a new notification, tap to view. Update the app to see the content right in the notification"; tapping it opens the event in the app as usual. If you really need older apps to keep showing notification content, turn on "Send plaintext pushes to older apps" under Settings → Notifications (the content is then relayed through Yunkan's servers, so it is not recommended long-term).
- A wrong device clock no longer costs you notifications. Many NAS boxes do not sync time over the network and drift by minutes. Notifications are now automatically aligned with server time, so clock drift can no longer silently stop phone notifications.
- Tell family, regular visitors and strangers apart even when the face is not visible. For doorway cameras that only see people's backs as they leave, or overhead cameras that only see the top of a head, each visit is now matched using that day's clothing, relative height and build, usual times and directions of coming and going, licence plates and similar clues. A conclusion is drawn only when there are enough reliable clues; otherwise it says "not sure" rather than guess. On a person's page you can set their household role and see how the system recognises them; event and visit cards show the identity and the evidence behind it, with 👍 / 👎 to correct it; automations can be conditioned on "family / stranger / a specific person", and notification templates gain identity variables. Available on the web, Android and iOS.
- Detection settings, simplified. More than a hundred parameters are now organised into 25 capability cards, each with an on/off switch, a one-line description, a Low / Standard / High sensitivity level and a few common options; turning on a capability automatically turns on what it depends on. Every detailed parameter is still there under "Expert mode" in the top-right corner. Redesigned on the web, Android and iOS.
- Name an unassigned face as a new person directly. No more uploading a photo under "Add person" first: choose "New person…" on an unassigned face, or select several and use "Name as new person", and the profile is created with the first face as its picture. Available on the web, Android and iOS.
- Choose Small, Medium or Large text in the web interface. The default is now one step larger than before; switch under Settings → Appearance → Text size, or Personalisation → Text size on the desktop. Small matches the previous size.
- Screenshot button in Playback (web, Android and iOS) saves the frame at the timeline cursor at full resolution. On phones the playback controls now sit on two rows: play, skip back/forward and previous/next event large and centred on top, with speed, screenshot and export on a smaller row below.
- On phones, switching Playback to a past day now starts playing from the beginning of that day, matching the web, instead of stopping at the end of the day with the play button greyed out.
- Clear the annotation queue in one click on the Training page. After a model update, press "Clear queue" to start fresh (optionally for one camera only); events, recordings and notification history are all kept, and new detections keep flowing in as usual. The queue now shows the true number of items waiting.
- No more missing recording segments. Occasionally a recording segment was not registered because of a restart or a busy system; it showed up as a gap in playback and was later deleted as junk. Those intact segments are now recovered automatically, and the recording health check now spots "continuous recording with gaps" and alerts you.
- Fixed snapshots filling the disk. The event snapshot folder could pile up large numbers of old images no event used any more; they could not be cleaned up and could even cause normal events to be deleted to free space. These unused images are now identified and removed automatically.
- Generic ONVIF cameras now get event snapshots. Cameras connected through ONVIF event subscription (including TP-Link VIGI) had no images in the event list or in notifications; this is fixed.
- Fixed "Enrolment failed" for face and pet photos on some NAS models. On a few NAS devices (such as some fnOS models) photo uploads failed while everything else worked; this is fixed.
- New automation templates: "Notify when someone appears" and "Notify on people, vehicles and animals". No zone drawing needed, all cameras by default. If you deleted the default notification rule, you can now recreate it from a template in one click.
- Line-crossing (direction) events added to Event priority and Subscribed events. Previously, once you picked specific subscription types, line-crossing notifications were silently filtered out; they can now be selected and prioritised on their own.
- Default automation rule names follow the interface language. English, French and Japanese setups no longer show a row of Chinese rule names; rules you have renamed are left untouched.
- The desktop "landing page" preference now really applies. Opening Yunkan from a bookmark takes you to the interface you chose instead of always the classic console, and desktop widget positions are remembered per screen size, so switching screens no longer stacks them in a pile.
- Faster sign-in on Android. When the server has both IPv4 and IPv6 addresses and one of them does not work, the Android app now tries both at once and uses whichever answers first, instead of waiting for a timeout.
- New installs use a lighter stream for detection by default, lowering CPU usage. Existing devices keep their current setting after the update.
- Install on a Mac. A new one-step Mac installer sets up a Linux virtual machine bridged to your local network and runs Yunkan inside it, so automatic camera discovery, app discovery on the local network, HomeKit and the rest work exactly as on a Linux host, and it starts automatically at boot.
- Raspberry Pi 5, ARM servers and other ARM hosts can install directly. The Standard (CPU) image is now available for both x86 and ARM, and the one-step installer picks the right one.
- Playback time no longer sticks at 00:00 when you drag back past midnight (web); Android playback dates follow the interface language, so French and other interfaces no longer show Chinese dates; long sidebar items in French and Japanese no longer squeeze their icons out.
- v0.9.53Highlight
Stronger sign-in: two-factor, passkeys and single sign-on · Sign in by scanning a QR code or approving on your phone · Arrange the mobile home screen your way
- Add a second step to sign-in. Under My Account → Sign-in security, link an authenticator app (Google Authenticator, Microsoft Authenticator, 1Password and the like); from then on you enter a 6-digit code after your password. Linking gives you 10 single-use recovery codes so a lost phone does not lock you out. Administrators can reset another user's two-factor setup from User management, and the on-device password-reset command gained a
--clear-mfaoption as the last resort for a sole administrator who has lost the recovery codes too. - Passkeys on the web. Sign in with your fingerprint, face or device password, either on its own or as the second step. Passkeys require access over an HTTPS domain, and the mobile apps do not support them yet.
- Single sign-on (SSO). If your home or office runs a standards-based identity service such as Authentik, Keycloak or Microsoft accounts, connect it under Settings → Single sign-on and an SSO button appears on the sign-in page. First sign-in creates the account automatically, or you can link it to an existing one. Supported on the web and in the Android and iOS apps. Three new guides on the website walk through two-factor, passkeys and SSO.
- Sign in by scanning a QR code. The web sign-in page has a new "Scan to sign in" tab: tap the scan button in the mobile app's top bar, confirm on the phone, and you are in without typing a password. The confirmation screen shows which device is signing in, where from, and whether it is on the same network as your phone; tap "Not me" and the account locks for 15 minutes.
- Approve sign-in on your phone. For accounts with two-factor enabled, choose "Approve on phone" after your password: the phone gets a push, you pick the number shown on the web page from three options, and sign-in completes.
- Integrations no longer need your password. A new authorisation page lets integrations such as Home Assistant send you to SkyView to sign in and tap "Authorise" to receive a dedicated token. API tokens can now be issued by every account, and administrators can list and revoke them per user. The matching Home Assistant integration release follows shortly.
- Arrange the mobile app's home screen your way (Android and iOS). Long-press a camera to enter edit mode, drag to reorder, tap to change size, tuck rarely watched cameras into the bottom tray, and cycle 1 / 2 / 3 columns from the top bar. The layout is saved to your account, so it follows you to a new phone.
- Mobile sign-in page redesigned (Android and iOS). Server details fold into one line and only expand when you tap Change; HTTP or HTTPS is inferred from what you type instead of a switch; two-factor and single sign-on both complete on the phone; failed sign-ins now explain the actual reason rather than a bare "401".
- Detection boxes and pose skeletons on notification images can be turned off separately. Two new switches under Settings → Notifications, both on by default, apply to event images sent to notification channels, webhooks and Home Assistant. Phone push images never carried overlays and are unaffected.
- Fall detection: objects in front of the lens are no longer mistaken for a person lying down. A customer had a white ball a few centimetres in front of a camera that kept being reported as a fall. A "lying" pose made up of only two or three scattered keypoints in a horizontal line no longer triggers a fall on its own; when there is hard evidence such as a collapse in body height, a downward motion or the head below the feet, alerts fire as before, so genuine falls are unaffected.
- Sign-in confirmation shows the real origin when accessed through the public single port. Previously, browsers arriving through the public single-port setup showed an empty origin address and no same-network indicator.
- Easier to investigate "there was a car / person but no event". The recognition score and threshold at the end of each tracked object are now written to the default log, so there is no need to switch to debug logging.
- Consistent wording in French, Japanese and Traditional Chinese. Terms such as tripwire, quiet hours and detection zone had two or three different names across the web, Android and iOS; they are now unified and match the website documentation.
- Add a second step to sign-in. Under My Account → Sign-in security, link an authenticator app (Google Authenticator, Microsoft Authenticator, 1Password and the like); from then on you enter a 6-digit code after your password. Linking gives you 10 single-use recovery codes so a lost phone does not lock you out. Administrators can reset another user's two-factor setup from User management, and the on-device password-reset command gained a